Polymers

Polymers
Properties and Applications 2
Editorial Board:
Prof. Dr. Hans-Joachim Cantow
Institut fiir Makromolekulare Chemie der Universitat
Stefan-Meier-StraBe 31, 7800 Freiburg/Germany
Prof. Dr. H. James Harwood
Institute of Polymer Science, University of Akron
Akron, OH 44325 / USA
Prof. Dr. Joseph P. Kennedy
Institute of Polymer Science, University of Akron
Akron, OH 44325 / USA
Prof. Dr. Anthony Ledwith
Dept. of Inorganic, Physical and Industrial Chemistry
Donnan Laboratories
Grove Street, P.O. Box 147, Liverpool L69 3BX / UK
Prof. Dr. Joachim Meif3ner
Techn.-Chem. Lab., Eidgenossische Techn. Hochschule
UniversitatsstraBe 6, CH-8006 Ziirich, Switzerland
Prof. Dr. Seizo Okamura
Professor of General Education Kyoto Sangyo University
Motoyama-Cho, Kamigamo,
Kita-Ku, 603 Kyoto, Japan
Dr. G. Olive I Prof. Dr. S. Olive
Monsanto Triangle Park, Development Center, Inc.
P.O.B. 12274, Research Triangle Park, NC 27709/ USA
H. H. Kausch
Polymer
Fracture
With 180 Figures
Springer-Verlag
Berlin Heidelberg New York 1978
Professor Dr. rer. nat.
HANS-HENNING KAuscH-BLECKEN VON SCHMELING
Laboratoire des Polymeres, Departement des Materiaux,

Ecole Polyteclmique Federale de Lausanne, SUisse
This volume continues the series Chemie, Physik und Technologie der
Kunststoffe in Einzeldarstellungen, which is now entitled Polymers/
Properties and Applications.
ISBN-13: 978-3-642-96462-6 e-ISBN-13: 978-3-642-96460-2

001: 10.1007/978-3-642-96460-2
Library of Congress Cataloging in Publication Data:

Kausch, H. H. Polymer fracture. (Polymers; v. 2). Bibliography: p. Includes index.
1. Polymers and polymerization - - Fracture. I. Title.
II. Series. Ta455.P58K38 620.1'92 78-14205
This work is subject to copyright. All rights are reserved, whether the whole or part
of the material is concerned, specifically those of translation, reprinting, re-use of
illustrations, broadcasting, reproduction by photocopying machine or similar means,
and storage in data banks. Under § 54 of the German Copyright Law where copies
are made for other than private use, a fee is payable to the publisher, the amount of
the fee to be determined by agreement with the publisher.
© by Springer-Verlag Berlin Heidelberg 1978

Softcover reprint of the hardcover 1st edition 1978
The use of registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the
relevant protective laws and regulations and therefore free for general use.
2152/3"020-543210
Preface
This book on "Polymer Fracture" might as well have been called "Kinetic Theory
of Polymer Fracture". The term "kinetic theory", however, needs some de-
finition or, at least, some explanation. A kinetic theory deals with and particu-
larly considers the effect of the existence and discrete size, of the motion and of
the physical properties of molecules on the macroscopic behavior of an ensemble,
gaseous or other. A kinetic theory of strength does have to consider additional
aspects such as elastic and anelastic deformations, chemical and physical reactions,
and the sequence and distribution of different disintegration steps.
In the last fifteen years considerable progress has been made in the latter do-
mains. The deformation and rupture of molecular chains, crystals, and morphologi-
cal structures have been intensively investigated. The understanding of the effect of
those processes on the strength of polymeric materials has especially been furthered
by the development and application of spectroscopical methods (ESR, IR) and of
the tools offracture mechanics. It is the aim of this book to relate the conventional
and successful statistical, parametrical, and continuum mechanical treatment of
fracture phenomena to new results on the behavior of highly stressed molecular
chains.
The ultimate deformation of chains, the kinetics of the mechanical formation
of free radicals and of their reactions, the initiation and propagation of crazes and
cracks, and the interpretation of fracture toughness and energy release rates in
terms of molecular parameters will be discussed. Although the prime interest had
been given to these subjects it was not possible to cover the literature completely.
The author apologizes in advance for all intentional and unintentional omissions
that will have occurred. In any event this book draws on the existing literature on
morphology, viscoelasticity, deformation and fracture of high polymers. It is hoped
that it forms a useful supplement of such monographs in view of a molecular inter-
pretation of polymer fracture.
Lausanne, September 1978 H. H. Kausch

Acknowledgements
The author would like to thank his friends, colleagues, and collaborators for their
encouragement and help. Their mUltiple efforts are highly esteemed. E. H. Andrews,
W. 0611, H. Hahn, D. Langbein, J. A. Sauer, W. O. Statton, J. Sohma, I. M. Ward, and
J. G. Williams have been kind enough to read appropriate chapters of this manuscript
and to make well appreciated comments. Many colleagues have assisted the timely pre-
paration of this book by making available copies of their relevant manuscripts prior to
publication. Thanks are also due to the many authors and publishers for their permissions
to reproduce graphic material and to K. L. DeVries, and W. O. Statton, Salt Lake
City, W. 0611, Freiburg, E. W. Fischer, Mainz, K. Friedrich, Bochum, E. Gaube,
Hoechst AG, S. Gogolewski, Lodz, J. W. S. Hearle, Manchester, H. Hendus, BASF AG,
D. Hull, Liverpool, G. Kampf, Bayer AG, P. Stockmayer, Stuttgart, and J. Wristers,
Baytown, for providing valuable photographs or electron micrographs.
The author acknowledges the financial assistance of the Deutsche Forschungs-
gemeinschaft during some of the studies carried out; he thanks the Battelle Institut
e. V., Frankfurt, which had made available its technical facilities during the early
stages of preparing the material; in organizing the manuscript the author has been
greatly assisted by his collaborators, Ing. EPFL T. Pesek and Dr. S. Sikka, and by the
speedy and reliable typing of Frau Weinhold and Mme Steireif. All this help is grate-
fully acknowledged.
Table of Contents
Chapter 1 Deformation and Fracture of High Polymers,

Defmition and Scope of Treatment 1
References 13
Chapter 2 Structnre and Deformation 15

I. Elements of the Superstructure of Solid Polymers. 15
A. Amorphous Regions 15
B. Crystallites 17
C. Superstructure 18
D. Characterization. 23
II. Deformation . 24
A. Phenomenology . 24
B. Molecular Description . 27
III. Model Representation of Deformation 29
References 37
Chapter 3 Statistical, Continuum Mechanical, and Rate Process

Theories of Fracture 40
I. Introduction . 40
II. Statistical Aspects 41
III. Continuum and Fracture Mechanics Approach . 47
A. Classical Failure Criteria 47
B. Fracture Mechanics . 50
C. Continuous Viscoelastic Models. 52
IV. Rate Process Theories of Fracture 53
A. Overview 53
B. Eyring' Theory of Flow 54
C. Tobolsky-Eyring . 55
D. Zhurkov, Bueche 56
E. Hsiao-Kausch. 59
F. Gotlib, Dobrodumov et al. 64
G. Bueche-Halpin 65
References 66
Chapter 4 Strength of Primary Bonds 69

I. Covalent Bonds . 69
A. Atomic Orbitals . 69
B. Hybridization 71
C. Molecular Orbitals 72
D. Multiple Bonds 73
VII
Table of Contents
II. Bond Energies . . . . . . . . . . . . . . . 73

A. Electronic Energy and Heat of Formation . . . . . 73
B. Binding Energy of Exited States and Radicalized Chains 80
1. Electronic Excitation 81
2. Ionization. . . . 82
3. Radicalization 84
III. Form of Binding Potential 86
References . . . . . 86
Chapter 5 Mechanical Excitation and Scission of a Chain 87

I. Stress-Strain Curve of a Single Chain . 87
A. Entropy Elastic Deformation 88
B. Energy Elastic Chain Deformation . 93
C. Non-Equilibrium Response . . . 98
II. Axial Mechanical Excitation of Chains 99
A. Secondary or van der Waal's Bonds 99
B. Static Displacements of Chains Against Crystal Lattices . 99
C. Thermally Activated Displacements of Chains Against Crystal
Lattices . . . . . . . . . . . . . . . . . 105
D. Chain Displacements Against Randomly Distributed Forces 107
E. Dynamic loading of a chain 108
III. Deexcitation of Chains 111
References 117
Chapter 6 Identification of ESR Spectra of Mechanically Formed Free

Radicals. . . 119
I. Formation. . 119
II. EPR Technique 120
A. Principles . 120
B. Hyperfine Structure of ESR Spectra 121
C. Number of Spins . . . . . . 122
III. Reactions and Means of Identification 123
IV. Assignment of Spectra. . . . . . 125
A. Free Radicals in Ground High Polymers 125
B. Free Radicals in Tensile Specimens 126
References . . . . . . . . . . 136
Chapter 7 Phenomenology of Free Radical Formation and of Relevant

Radical Reactions (Dependence on Strain, Time, and Sample
Treatment) . . . . . . . . . . . . . 141
I. Radical Formation in Thermoplastics. . . . . 141
A. Constant Rate and Stepwise Loading of Fibers . 141
B. Effect of Strain Rate on Radical Production 150
C. Effect of Temperature. . . . . 152
1. Apparent Energy of Bond Scission 152
2. Rate of Bond Scission . . . 154
VIII
Table of Contents
3. Concentration at Break 156

D. Effect of Sample Treatment 159
II. Free Radicals in Stressed Rubbers 162
A. Preorientation, Ductility, and Chain Scission 162
B. Cross-Link Density, Impurities, Fillers 165
III. Mechanically Relevant Radical Reactions 167
A. Transfer Reactions. . . 168
B. Recombination and Decay 169
C. Anomalous Decay . . 170
D. Radical Trapping Sites . 170
References .... 171
Chapter 8 The Role of Chain Scission in Homogenous

Deformation and Fracture . . . . . . . . . . . 173
I. Small-Strain Deformation and Fracture of Highly Oriented 174
Polymers . . . ....
A. Underlying Problems . . . . . . . . . . . 174
B. Loading of Chains before Scission. . . . . 175
C. Spatially Homogeneously Distributed Chain Scissions 183
D. Formation of Microcracks 193
E. Energy Release in Chain Scission 196
F. Fatigue Fracture of Fibers 198
G. Fractography. . . . . . . 200
II. Deformation, Creep, and Fatigue of Unoriented Polymers 204
A. Impact Loading. . . . . 204
B. Failure under Constant Load 211
C. Homogeneous Fatigue. . . 220
1. Phenomenology and Experimental Parameters 220
2. Thermal Fatigue Failure . . . . . . . 222
3. Wohler Curves . . . . 223
4. Molecular Interpretations of Polymer Fatigue 224
D. Yielding, Necking, Drawing 230
E. Elastomers . . . . 235
III. Environmental Degradation 237
References . . . . . 243
Chapter 9 Molecular Chains in Heterogeneous Fracture 251

I. Fracture Mechanics. . . . 252
A. Stress Concentration 252
B. Sub critical Crack Growth. 259
C. Critical Energy Release Rates 271
II. Crazing. . . . 272
A. Phenomenology. . . . . 272
B. Craze Initiation . . . . . 276
C. Molecular Interpretation of Craze Propagation and
Breakdown 286
D. Response to Environment 290
IX
Table of Contents
III. Molecular and Morphological Aspects in Crack Propagation 293

A. Fracture Surfaces . . . . . . . 293
B. Notched Tensile and Impact Fracture. 306
C. Fatigue Cracks . . 310
D. Mechano"Chemistry . . . . . . 312
References 315
Appendix Table A-I. List of Abbrevations of the Most Important Polymers 321
Table A-2. List of Abbrevations not Referring to Polymer Names 322
Table A-3. List of Symbols. . 323
Table A-4. Conversion Factors. 325
Subject Index. . . . . . . . . . . 326
x
Chapter 1
Deformation and Fracture of High Polymers,

Definition and Scope of Treatment
The importance of a thorough understanding of the deformation behavior and the

strength of polymeric engineering materials need not be emphasized. It is obvious
to anyone who wants to use polymers as load bearing, weather-resisting, or deform-
able components or who wants to grind or degrade them. "Strength" and "frac-
ture" of a sample are the positive and negative aspects of one and the same phenom-
enon, namely that of stress-biased material disintegration. The final step of such
disintegration manifests itself as macroscopic failure of the component under use,
be it a water pipe, a glass fiber reinforced oil tank, or a plastic grocery bag. The
preceding intermediate steps - nonlinear deformation, environmental attack, and
crack initiation and growth - are often less obvious, although they cause and/or
constitute the damage developed within a loaded sample.
Efforts of fracture research are directed, therefore, towards three goals:
1. To understand the cause and development of a failure and to assess the im-
portance of anyone process contributing to a failure or delaying it
2. To see whether, how, and with what effect the most damaging process can
be controlled or even eliminated, and how the strength-determining material prop-
erties can be improved
3. With the first two goals only partly attainable, one would like to have know-
ledge of the endurance limit of a part in terms of stress or strain level, environment,
number of loading cycles, or of loading time.
The first goal is equivalent to the ability to predict the material response under
all kinds of foreseeable conditions of mechanical, thermal, or environmental attack,
to characterize the constantly changing state of the material, and to know at any
moment which conditions of attack are critical for the state the material just at-
tained. To achieve this goal in all generality is hardly possible, in view of the com-
plexity involved in characterizing polymeric structures, let alone their deformation.
A conclusive discussion of those two subjects and of their interrelation with and
dependency on other aspects of polymer science is perhaps achieved in the frame-
work of polymer science encyclopedias [1,2] and partially in specialized textbooks
on the viscoelastic [3, 4], mechanical [5], and physical [6-8] behavior of polymers.
Although the first goal cannot be achieved in all generality, the literature on frac-
ture (cf. textbooks [9-12]) deals extenSively and successfully with interpretations
of the fracture event which focus on just those processes which account for the
most obvious changes of sample morphology during fracture development.
1. Deformation and Fracture of High Polymers, Definition and Scope of Treatment
The previous statements on the goals of fracture research are well illustrated
by the fracture patterns of rigid polyvinylchloride (PVC) shown in Figures
1.1-1.3. The samples, water pipes internally pressurized, fractured in a brittle man-
ner when the circumferential stress level was high, in a partially ductile manner
and only after prolonged times at medium stresses, and through thermal crack
growth (creep crazing) after very long times at low stress levels. The three processes
leading to pipe failure in these three cases are respectively the rapid extension of a
flaw, the flow of matter, and the thermally activated growth of a flaw. In all three
cases the volume element within which the fracture is initiated is finite, consequent-
ly nonhomogeneous deformation must have occurred locally. We will deal with the
nature and interpretation of such heterogeneous deformation of a supposedly homo-
geneous material later.
Fig. 1.1. Brittle fracture of an in- fig. 1. 2. Ductile failure of a PVC pipe at medium stress
ternally pressurized water pipe levels (rupture as a result of the reduction of wall thick-
(polyvinylchloride, PVC), taken ness at a defect at the indicated position) , taken from
from Ref. [13]. Ref. [131.
Fig. 1. 3. Long-time failure of a PVC pipe through

development of a creep craze (from [l3]).
We also understand the importance of the first goal with respect to achieving
the second. In the described example of pressurized PVC pipes it was known that
in the medium time range the performance of a pipe was limited by its creep be-
havior. We would wish, therefore, to decrease the creep rate by increasing the rigidi-
ty of the material; in doeing so, however, we would facilitate crack propagation
and, possibly, craze initiation and thus reduce the permissible stress levels at short
and very long times of application.
2
The third goal, the knowledge of the endurance limit of a part in terms of stress
or strain level, is generally directly attainable within a certain experimental region.
Figure 1.4 summarizes, for instance, the test results on PVC pipes at temperatures
between 20 and 60°C, circumferential stresses between 10 and 50 MN/m 2 and a
range of times to fracture of 0.3 to 104 h. Such representations, which in nongraphi-
50,r-~~-------'--------r---'
MN
m2
I~~-+-~
I
I
~
o
~ 30 -
I I
----t----j
Vi i I
c::""
+
I
.;: I
i
1
C I
201t----t~~ ----+--1---11
~
::>
·u
"'~.•
• I II I
I
.01- 1 I
101r--+----~I---~·~~i~4~1
° 20°C 004 brittle fracture
o 40°C •• '" ductile fracture
'" 60 0 C • • • creep craz ing
Fig. 1.4. Times-to-failure of internally
pressurized PVC water pipes at different
ti me to fracture stresses and temperatures (after 113]).
cal form can be extended to interrelations between more than three parameters,
form an important basis of our understanding of material behavior within the multi-
dimensional space covered by the parameters investigated. Material selection and
optimization of its properties through parameter variation become possible to the
extent the mathematical framework of parameter-property coefficients holds. With
regard to fracture this means that one would like to define a fracture criterion in
terms of the relevant structural and environmental parameters.
A good example for the occurrence of unpredicted events is offered by the
behavior of polyethylene (PE) pipes - and of other semicrystalline materials -
under constant circumferential stress; they showed a rapid drop of sustained stresses
after long loading times (Fig. 1.5). With PE - as with PVC - an initial time range is
observed where the times to failure are only a weak function of stress. Depending
on the temperature, the failure mode is either brittle (Fig. 1.1) or ductile (Fig. 1.2
and 1.6). Both materials are also comparable insofar as the thermally activated
growth of a creep crack (Fig. 1.3, 1.7, and 1.8) can be held responsible for the fail-
3
20
In
In
G/
In
MN -"':8
....... JTi210 -..:." - o·
~ i'>- ___ p 0 0
8
"
.-J
d ~ic
:;:; 6 .j
'tJ
'"
1"1
"'-..
-...
C
G/ 'V
4 ~ "
G/
•• ~ ~
E
:3
u \ [\
'.~
u
2 f- o'=p ~~
"b
f- II II I
.
10 10 2 10 3 10 4 10 5 10e hours
1-......., ""'I............
' oil 2
, 10 10 year s
time to fracture
Fig. 1.5. Times-to-failure of high-density polyethylene (HOPE) water pipes under internal
pressure p at different stresses and temperatures. d m : average diameter , s: wall thickness ; A:
ductile failure, B: creep crazing (specimens are shown in Figs. 1.6-1.8; after 114]).
Fig. 1.6. Ductile failure of a HOPE water pipe under

the conditions shown as point A in Fig. 1.5:
u v =7 MN/ m2 , T= SO °C(aftci[14])
Fig. 1.7. Surface of a creep craze formed in Fig. I.S. Detail of the fracture surface
HOPE under conditions shown as point B in close to the upper center of the mirror
fig. 1.5 : U v = 6 MN/m2, T= SO °C (after [14]). zone, the point of craze initiation
(after 114]).
4
ure of a pipe after long times of service. Both materials are different insofar, as one
finds in PE - but practically not in PVC - that the kinetics of creep crack develop-
ment differs markedly from that of ductile failure (Fig. 1.5). This finding only under-
lines that it is necessary to deal directly with the physical nature of material defect
development in order to predict reliably material behavior - especially for new ap-
plications - and/or improve properties through additional components or modified
production.
In this monograph we wish to discuss the physical nature of defect develop-
ment, and we will focus on linear thermoplastic and elastomeric polymers (Table 1.1.).
We recognize that these materials cover a wide range of properties, although they
are composed of molecules which exhibit a number of similarities: they are pre-
dominantly linear, flexible, highly anisotropic (nonp.xtended) chains having molecu-
lar weights between 20000 and more than 1000000. Figure 1.9 gives a represen-
.........__.::;;;..--4-- molecule
- - - - chain end-to-
end distance
bond +---- segment

vectors
Fig. 1.9. Representation of

linear flexible chain molecule
(polyamide 6), (after [21]).
tation of a (polyamide 6) chain showing the nonextended collformation, an

arbitrary division into segments, and - in the insert - the primary bonds forming
the backbone chain The relative sizes of the atoms and the proportion of volume
occupied by them within the chain is illustrated by the Stuart model of a polyamide
segment (Fig. 1.10). In actual size this segment has an extended length of 1.97 nm.
If such a segment could be stressed in the chain axis direction, the bending and
stretching of primary bonds would result in a chain stiffness of about 200 GN/m 2
[15], whereas the intermolecular attraction of segments through the much weaker
van der Waal's forces entails a stiffness of "only" 3 to 8 GN/m 2 in a direction per-
pendicular to the chain axis. The fact that polymer molecules are long, anisotropic,
and flexible chains accounts for the characteristic properties of polymeric solids,
5
0\ Table 1.1. Mechanical Properties of Chain Molecules and Polymeric Solids*) ~
0
G
Isotropic Solid
8'
..,
Chain molecule Oriented Fiber !17]
~o·
Polymer Abbrev. Monomer Monomer Young's Young's Strength Young's Strength Critical :;
(lSO/R molecular length modulus modulus modulus stress §
1043- weight L(10- lOm) Ec(MN/m2) E E ab 0-
ab ac
",
-1969) Mo (from unit (Refs. [15, 16]) (MN/m 2) (MN/m2) (MN/m2) (MN/m2) (MN/m2) ..,
(g/mol) cell) (Ref. [18]) '"
g
(!)
0
....,
Semicrystalline Polymers :r:
QQ'
~
1. Polyethylene PE 28 235,000 "'0 "

high density 2.534 8000-50.000 480 1000 27 160 -<
;3
low density 2.54 [19] 200 13
2. Polypropylene PP 42 6.5/3 41,000 6000-9000 1250-2400 33 98 ~~
0
3. Polytetratluorethylcnc PTFE 100 2.60 153,000 400-980 23 117 G
4. Poly(oxymethylene) POM 1.92

::n
:;
30 53,000 1200-2550 3500 68 216 :::;:
5. Poly(ethylene oxide) (PEO) 44 19.3/9 9800 530 o·
:;
6. Poly(propylene oxide) 58 7.15/2 133
:;
7. Polyethylene tereph- 192 10.77 74,000 137,000 20,000-25,000 800-1400 3000 53 155 '0-"
thala!e PETP VJ
8. Polyamide 8
'tl
(!)
polycaprolactam PA 6 226 17.2 24,500 5000-8500 390-870 1250 61 0
....,
polyhexamcthylene
adipamidc PA 66 226 17.3 420-930 1500-4100 80 179
.......,
G
aromatic ~
(Kevlar 49) 119 6.6 180,000 144,000 3000 S(1)
a
Amorphous Polymers
9. Polystyrene PS 104 2.21 9000-11 ,800 3250 50

10. Polyvinyl chloride PVC 62.5 2.55 (160,000-230,000) 2500-3000 55 142
11. Polyvinyl alcohol PVAL 44 2.52 250,000 35,000-50,000 1200-2800 28
Table 1.1. (continued)
Chain molecule Oriented Fiber [17] Isotropic Solid
Polymer Abbrev. Monomer Monomer Young's Young's Strength Young's Strength Critical ~
(ISO/R molecular length modulus modulus modulus stress ~
(D
1043- weight L(lO-lOm) Ec(MN/m2) E ub E ub Uc 8'

-1969) Mo (from unit (Ref~. [15, 161) (MN/m2) (MN/m2) (MN/m2) (MN/m2) (MN/m2)
(g/mol) cell (Ref. (18»
::s
i·
g
Po.
Amorphous Polymers
~
12. Polyvinyl acetate PVAC 86 1300-2800 29-49 a
13. Polymethyl meth- ~
o
...,
acrylate PMMA 100 2.11 3000-4000 70 68
14. Polycarbonate ::c
QQ'
(bisphenol A) PC 254 10.75 2350 53 145 ::r
~
'<
i3
(Cross-linked Rubbers, ~
Y'
for comparison) ~
Natural rubber NR 136 8.1 1-2 17-25

?g:
Butyl rubber IlR 56 4.7 0.7-1.5 18-21 .,::s
Polyurethane PUR 0.50 20-30 6.
Silicon rubber SIR 0.15 2-8 en
8
't:l
(D
*) Reference values of technical polymers at room temperature (critical stresses refer to temperature of ductile-brittle transition); data taken from o
...,
D. W. van Krevelen [7 J and Polymer Handbook [20 J unless otherwise indicated. ::;l
~
i3
<l>
-.l :a
Fig. 1.10. Arbitrary segment of polyamide 6

chain (extended length 1.97 nm, cross
section 0.176 nm 2 ).
namely the anisotropy of macroscopic properties, the microscopic heterogeneity

and the nonlinearity and strong time-dependency of polymer response to mechanical
excitation.
All solid plastics, from rubbery networks to rigid glasslike bodies, are aggregates
of such molecular chains, which cohere through physical (temporary) or chemical
(permanent) "cross-links". The wide variety of the macroscopic mechanical properties
of these materials is caused by differences in aggregation (crystallinity or super-
structure) and in molecular parameters, as for instance by differences in chain
structure, intermolecular segmental attraction , segment mobility, or cross-link den-
sity. Unfortunately these parameters are not fixed values attached to a molecular
species or a polymeric body, but depend on polymer processing and on environ-
mental conditions. In fact simply by changing the time or temperature scale of
the mechanical excitation of one and the same polymer a wide range of responses
will be obtained: from that of a brittle glass to that of a highly viscous liquid.
This general behavior is depicted in a time-temperature diagram, given by Retting
[22] for hard PVC (Fig. 1.11). Within the relatively small temperature interval
from 0 to 100 °C (rapid loading) or - 100 to +60 °c (creep) the indicated changes
from glass to rubber, from - macroscopically - brittle to highly ductile defor-
s
I
, Main ..
Mai n rel a xatio/n t ran sit io n
MELT ma x imum
I
7
100
Medium ex tensibi l ity /
..
E High extensibility
o medium-low rupture stress Sec ond!lry
ru pture of semi
.
relaxation maximum
~
low rup t ure tough t ype ,
""7i
stress, rupture
0.01
lIT
I! ! I!! !!! ,
200 100 50 o -50 e

te mpera ture
Fig. 1.11. Time-temperatur e diagram of the mechanical properties of PVC (after Retting [22 D.
8
mation and fracture are observed. The changes in the type of response of a given
aggregate of chains clearly indicate that different deformation processes must be
dominant in different regions of the time-temperature regime.
In discussing the physical nature of defect development it is necessary to speci-
fy which form of local deformation eventually leads to a disintegration of the poly-
meric solid or to a localized heterogeneous deformation and to crack initiation. In
previously proposed kinetic theories quite different processes were held responsible
for the development of macroscopic failure. Viscous flow (within cotton fibers) was
suggested by Busse and his colleagues [23] to determine the fatigue life of these
fibers. The "slippage" of secondary bonds and the net decrease of the number of
bonds was considered by Tobolsky and Eyring [24] for the two cases where repair
of bonds under stress is possible or impossible. A model involving two independent
networks was discussed by Stuart and Anderson [25] who argued that only fracture
within the first network with unsymmetrical potential barriers between broken and
unbroken states contributes to failure of glasses. A theory of correlated Brownian
motion of groups of atoms (VielJachplatzwechsel) proposed by Holzmiiller [26,
27] is based on Boltzmann statistics of phonon energy distribution where the pho-
nons are the independent statistical units.
The idea that the breaking of primary chemical bonds plays a decisive role in
fracture was independently forwarded by Zhurkov [28, 29] and Bueche [30, 31].
Bueche, and later Halpin [32, 33] and Smith [34, 35], had been concerned with the
strength of rubbery polymers, and found that the time-temperature dependence
of the ultimate properties could be explained by the deformation behavior and the
finite extensibility of molecular strands. Prevorsek and Lyons [36, 37] emphasized
that the random thermal motion of chain segments leads to nucleation and growth
of voids without breakage of any load-carrying bonds. The lifetime of a sample
then is determined by the time it takes one flaw to grow to a critical size. Niklas
and Kausch [13] discussed the effect of the dissociation of dipole-dipole associates
on the strength of PVC.
To the molecular processes specified above one has to add the unspecified ones:
internal destruction, damage, or crack formation probability. In analogy to the
molecular description of deformation used by Blasenbrey and Pechhold [38] all of
these molecular processes can be referred to the four physical rearrangements that
can occur between neighboring chain segments with parallel chain axes: change of
conformation (segment rotation, gauche-trans transformation), cavitation, slip, and
chain scission. In Figure 1.12 these rearrangements of segment pairs are shown. Of
these chain scission and - to some extent - cavitation and slip are potentially
detrimental to the load-bearing capacity of a polymer network. Conformational
changes on the other hand appear to be "conservative" processes, which by them-
selves will modify and delay but never generate any progress in the development
of a fracture process.
Here it seems to be necessary to say a word about the meaning of the terms
fracture, rupture, and failure as they will be used within this text. Andrews [9] says,
"Fracture is the creation of new surfaces within a body ... The term rupture may
be regarded as truly synonymous but not the term failure since the latter indicates
only that some change has taken place in the body (generally an engineering com-
9
transformation of pairs excess volume molecular process

of segments of pairs
1.
I~ - I~ 3
TVo
- II 0
change of conformation
-
(segment rotation, gauche-
trans or trans - gauche
~~
1 transformation)
T Vo
- Ii
J..
4 Vo
2.
I~ - 3
>2 Vo cavitation
3.
I~ - I~ 3
7: Vo slip
L.
I~ - :~ >t Vo chain scission
Fig. 1.12. Four basic transformations of pairs of linear chain segments. Excess volume of pairs
refers to PE, with Vo being the volume of a CH 2-group 1381.
ponent of some kind) which renders it unsuitable for its intended use. Fracture is
only one of many possible modes of failure". We will supplement this definition
by saying that as a result of the creation of new surfaces either holes have to open
up or formerly connected parts of the fractured specimen have to become detached
from each other.
In mathematical terms we may define a fracture process as one by which the
sum of the degrees of connectedness of the parts of a sample or a structural element
is increased by at least one. This definition includes the piercing of films and tubes,
but excludes minute damages as caused by crazing or chain scission.
The term fracture process is meant to describe the whole development of a
fracture from its initiation through crack growth and propagation to completion
of the fracture. In contrast to thermal or environmental degradation and failure
10
the fracture process is understood as stress-biased material disintegration. It carries

on if and only if a driving force exists, which has literally to be a force. Forces,how-
ever, will give rise to deformations. This means that fracture initiation will always
be preceeded by specimen deformation.
From Figure 1.12 it could be concluded that, in the absence of flow, chain scis-
sion and chain strength determine the mechanical properties of polymers. In fact,
there are a number of observations which support this assumption:
with increasing number n of chain atoms, a saturated linear hydrocarbon will
turn from gas to liqUid (n;> 5) and to solid (n ;> 18); the solid will attain tech-
nically useful strength (10 MN/m2 ) if n is around 2000; with increasing molecular
weight and chain orientation that strength may still be increased 50-fold [19,
20];
the mechanical loading and rupture of virtually all natural and synthetic polymers
leads to the scission of molecular chains and the formation of free radicals [39];
the strength and the hardness of solids increase with volume concentration of
primary bonds [40];
in a number of uniaxial tensile fracture experiments the energy of activation of
sample fracture coincided with the activation energy for main chain bond
scission [39].
The observation that sample strength increases with molecular weight and chain
orientation is illustrated in more detail by Figure 1.13. In low molecular weight ma-
terial chain slip occurs readily, so that the sample strength depends solely on the
strength of intermolecular attraction. Noticeable macroscopic strength is only ob-
tained once the molecular weight is sufficient to permit physical cross-links through
entanglements or chain folds between several chains. In the range of molecular
weights between 1.5 and 3 x 104 g/mol, fiber strength increases with molecular
weight, whereas at still higher molecular weights, the effect of the increasing num-
ber of defects introduced in processing such a fiber, tends to outweigh the effect
of the decreasing number of chain ends (Fig. 1.13, Curve I). It is also indicated in
Figure 1.13 (Curves I-IV) that high strength with a high, degree of orientation (e.g.,
obtained at higher drawing temperatures Td ) requires high molecular weights [41].
The effects of crystal structure and molecular weight on macroscopic tensile strength
are shown by curves V and VI [43].
On the other hand, the data of Table I do not support a simple relation between
the strength values of individual chains and of the solids formed thereof. Young's
moduli of bulk specimens are one to two orders of magnitude smaller than the
corresponding chain moduli, bulk strengths at room temperature are two to three
orders of magnitude smaller than chain strengths. The critical stress, which is the
fracture stress at that temperature where fracture changes from ductile to brittle,
i.e., at a temperature well below the glass transition temperature, is also one to two
orders of magnitude smaller than the corresponding chain strengths. The tensile
strengths observed at different morphological levels are indicated in Figure 1.14.
In the cited monographs [2-10] very little has been said about the details of
how molecular destruction processes effect ultimate properties. In the last 10 years,
however, considerable progress has been made in precisely this area, in the investi-
gation of the deformation, strength, and scission of chain molecules. Mainly dealing
11
~'"
~,
to "
/
..........
n Td=150'C~ '
.....
0.8 // x/, m 140 •C

1
i I;~
~ L/
I
I v Y monoclinic
I
I II / Ilr 20'C
I
I /1J/ I~"
lZl hexagonal
0.4
j
I I nm
,! ~ 300
.c.
,// g.
~ .!!
02 200 c
/' 'a
.c.
u
."./
100
,...'
/./
o /'
o 2 4 10' glmole M
molecular weight
Fig. 1.13. Effect of molecular weight and temperature Td of the fiber orientation process
(drawing) on the tensile strength of polyamide 6. I: low·dernier filaments, Td = 170°C [411,
II~IV: effect of drawing temperature, Td = 150°C, 140 °c, 20°C [421, V, VI: effect of crystal
structure [421.
with the notable contributions of Russian scientists working in this area, the work
by Regel, Slutsker, and Tomashevskii [11] has just become available in Russian.
With these words of introduction we are in a position better to define the scope
of this book. The central theme is the behavior of chain molecules under conditions
of extreme mechanical excitation and the likely role of chain scission in the initia-
tion and development of macroscopic fracture. We will study the structure and de-
formation of solid polymers (Chapter 2) in an attempt to define the conditions
under which elastic straining of chains occurs. We will then discuss existing non·
morphologic fracture theories (Chapter 3) in view of our central theme. The mo·
lecular description of fracture begins with a brief account of the strength of primary
bonds (Chapter 4) and of the thermo mechanical excitation and scission of chain
segments (Chapter 5). The study of free radicals formed as a consequence of stress·
induced chain scission (Chapters 6~8) constitutes the main part of this volume. It
is the only part of this book where a comprehensive coverage of the existing litera-
ture has been attempted. In the final chapter, 9, the role of molecular chains in
heterogeneous fracture will be analyzed. For this purpose a brief introduction is
given into the mechanics of fracture (stress analysis of cracks in elastic and elastic·
plastic materials, slow crack propagation). The fracture mechanical functions criti-
cal energy release rate GIc and fracture toughness KIc are interpreted from a non·
12
.References for Chapter 1
C -C bonds
MN/m 2
( 188kJ/rrole)
** 01s +
1 h +
chain segments
lOOOOf- (5nm)
~1000f • X = 59
iii oriented filaments
c
2 • X = 37
b. critical stress
non-oriented, spherulltlc
100f-
1 0.5-1% Hp
1 2 - 4% H20 + Perepelkm
* Kausch
• Levin et at.
x Zilvar
A Vincent
10~------------------------------------------~
Fig. 1.14. Tensile strengths of polyamide 6 at different levels of morphologic organization.
continuum viewpoint. The same approach is applied to the phenomenon of craz-

ing. A discussion of molecular and morphological aspects of crack propagation and
of possible failure criteria concludes this monograph.
As mentioned in the introduction, a kinetic theory of fracture deals with and
particularly considers the existence and discrete size, the motion, and the response
of molecules or their parts in a stressed sample. It thus requires a molecular de-
scription of polymer structure and deformation. Such a description is available in
the cited books [1-12], [20] or papers [15,38]. In Chapter 2 only a very brief
summary is offered which is not meant to be self-supporting or complete.
References for Chapter 1
1. H. A. Stuart (ed.): Physik der Hochpolymeren. Berlin - Heidelberg: Springer-Verlag 1952.

2. H. F. Mark, N. G. Gaylord, N. M. Bikales (eds.): Encyclopedia of Polymer Science and
Technology. New York: Interscience 1964.
3. J. D. Ferry: Viscoelastic Properties of Polymers (2nd ed.). New York - London: Wiley
1970.
4. N. G. McCrum, B. E. Read, G. Williams: Anelastic and Dielectric Effects in Polymeric
Solids. London - New York: Wiley 1967.
13
5. I. M. Ward: Mechanical Properties of Solid Polymers. London: Wiley-Inter science 1971.

6. A. D. Jenkins (ed.): Polymer Science. Amsterdam - London: North Holland 1972.
7. D. W. van Krevelen: Properties of Polymers, Correlations with Chemical Structures. Amster-
dam - London: Elsevier 1972.
8. J. A. Manson, L. H. Sperling: Polymer Blends and Composites. New York - London: Plenum
1976.
9. E. H. Andrews: Fracture in Polymers. Edinburgh London: American Elsevier 1968.
10. G. M. Bartenev, Yu. S. Zuyev: Strength and Failure of Viscoelastic Materials. Oxford - Lon-
don: Pergamon 1968.
11. V. R. Regel, A. 1. Slutsker, E. E. Tomashevskii: Kinetic Nature of the Strength of Solids
(in Russian). Moscow: Nauka 1974.
12. R. Hertzberg: Deformation and Fracture Mechanics of Engineering Materials. New York:
Wiley 1976.
l3. H. Niklas, H. H. Kausch: Kunststoffe 53, 839 and 886 (1963).
14. E. Gaube, H. H. Kausch: Kunststoffe 63, 391-397 (1973).
15. J. A. Sauer, A. E. Woodward, In: Polymer Thermal Analysis. Vol. II. P. E. Slade, L. T.
Jenkins (eds.). New York: Marcel Dekker 1970. pp. 107 -224.
16. I. Sakurada, T. Ito, K. Nakamae: J. Polym. Sci. 15, 75 (1966).
17. K. E. Perepelkin: Angew. Makrom. Chemie 22, 181-204 (1972).
18. P. I. Vincent: Polymer 13,558 (1972).
19. N. J. Capiati, R. S. Porter: J. Polym. Sci., Polym. Phys. Ed. 13, 11 77 -1186 (1975).
20. J. Brandrup, E. H. Immergut: Polymer Handbook (2nd ed.). New York: Wiley 1975.
21. H. H. Kausch: J. Polymer Sci. C32, 1-44 (1971).
22. W. Retting: Kolloid-Z 213, 69 (1966).
23. W. F. Busse, E. T. Lessing, D. L. Loughborough, L. Larrick: J. Appl. Phys.13, 715 (1942).
24. A. Tobolsky, H. Eyring: J. Chern. Phys.11, 125 (1943).
25. O. L. Anderson, D. A. Stuart: Ind. Eng. Chern. 46, 154 (1954).
26. W. Holzmiiller: Z. Phys. Chemie (Leipzig) 202,440 (1954) and 203,163 (1954).
27. W. Holzmiiller: Kolloid-Z 203, 7 (1965) and 205, 24 (1965).
28. S. N. Zhurkov, B. N. Narzulaev: J. Techn. Phys. (USSR) 23,1677 (1953).
29. S. N. Zhurkov: Z. Phys. Chemie (Leipzig) 213, 183 (1960).
30. F. Bueche: J. Appl. Phys. 26, 1133 (1955) and 28,784 (1957).
31. F. Bueche, J. C. Halpin: J. Appl. Phys. 35, 36 (1964).
32. J. C. Halpin: Rubber Chern. Technol. 38,1007 (1965).
33. J. C. Halpin, H. W. Polley: J. Composite Materials 1,64 (1967).
34. T. L. Smith: J. Polym. Sci. 32, 99 (1958).
35. T. L. Smith: J. Polym. Sci. A, 1, 3597 (1963).
36. D. C. Prevorsek, W. 1. C. Withwell: Textiles Res. 1. 33,963 (1963).
37. D. C. Prevorsek, M. L. Brooks: 1. Appl. Polym. Sci. 11, 925 (1967).
38. W. Pechhold, S. Blasenbrey: Kolloid-Z. Z. Polymere 241, 955 (1970).
39. S. N. Zhurkov: Int. 1. of Fracture Mechanics 1, 311-323 (1965).
40. L. Holliday, W. A. Holmes-Walker: J. Appl. Polym. Sci. 16, 139-155 (1972).
41. D. Prevorsek: 1. Polym. Sci. C32, 343-375 (1971).
42. V. M. Poludennaja, N. S. Volkova, D. N. Archangelskij, A. G. Zigockij, A. A. Konkin:
Chim. Volokna 5, 6-7 (1969)
43. H. Kraessig: Textilveredlung 4, 26- 37 (1969)
44. B. Ya. Levin, A. V. Savitskii, A. Ya. Savostin, E. Yeo Tomashevskii: Polymer Science
USSR 13, 1061-1068(1971).
45. V. Zilvar, 1. Boukal, 1. Hell: Trans. J. Plastics Inst. 35, 403-408 (1967).
14
Chapter 2
Structure and Deformation
I. Elements of the Superstructure of Solid Polymers . 15

A. Amorphous Regions 15
B. Crystallites . 17
C. Superstructure . 18
D. Characterization 23
II. Deformation 24
A. Phenomenology. 24
B. Molecular Description 27
III. Model Representation of Structure and Deformation 29
I. Elements of the Superstructure of Solid Polymers
A. Amorphous Regions
The basic structural elements of high polymer solids are the chain rr.olecules. The
variety of their structure and flexibility permits different modes of organization
and mechanical interaction. At this point the characteristic elements of structure
and superstructure of amorphous and semi crystalline polymers will be introduced.
The interrelations between chain parameters (structure and regularity), crystal or
superstructural parameters (degree of crystallinity, lattice structure, nucleation and
growth kinetics, defects), and environmental parameters are extensively discussed
in the literature and are not the object of this book [1-3].
If no regular organizatzion or no colloidal structure within a solid aggregate of
chains becomes apparent one speaks of an amorphous state. In recent years the
interpretation of the non-ordered or amorphous state of solid high polymers has
been subject to lively discussions and careful studies. Until about 1960 the general
view prevailed that the chain molecules in isotropic, non-crystalline polymers as in
many rubber, glassy polymers such as PS*), PVC, PMMA, PC, or quenched "semi-
crystalline" polymers such as PCTFE, PTFE, PETP have a random conformation
and that the random coil or spaghetti model gives an adequate description of the
"structure" of these polymers. In the years following 1960 the concept of a close
range order of the chain molecules within X-ray amorphous polymers gained grow-
* In the annex (Table A.l) a list of standardized abbreviations of those polymer names is given
which are used in this monograph. The abbreviations are in accordance with the international
standards proposed by ASTM. DIN, and ISO.
15
2. Structure and Deformation
ing support. The evidence for a close range order was thought to derive from a com-
parison of segment volume and amorphous density, from electron microscopic ob-
servation of structural elements, from calorimetric investigations, from a discussion
of crystallisation kinetics, and from a study of network orientation. After 1970
Kirste, Schelten, Fischer, Benoit, and their associates [4-6,91] were applying neu-
tron scattering techniques to amorphous polymers in addition to employing light
and small angle x-ray scattering and spectroscopic methods. They concluded that
their experiments and the previous experiments of other authors have given no
positive clue either against the random coil hypothesis or in favor of the concept
of an existing close range order in PS, PMMA, PC, SI or the amorphous regions of
PE. Flory [7], [8] also summarizes his extended observations on polymers in solu-
tion, rubbery networks, and amorphous one- and two-phase materials in favor of
the random coil structure of chains in bulk. These authors [4-8], [9], [100 - 104 ]
state in fact that the radius of gyration of a polymer molecule in the bulk agrees
well with the unperturbed dimensions measured in the melt or in dilute solution.
Hosemann, Pechhold, and Yeh [9 - 12] propose that amorphous states and
defect-rich, Originally ordered states are commensurable. They thus retain elements
a c
,- I
I I
,"" ...... -
I I
./
,
.
... ..,- - --; ..
I
.... r .... I
J ... .... 1
d I I
e
Fig. 2.la-e. Model representations of the amorphous state. (a) interpenetrating coils [4, 71,
(b) and (c) honeycomb and meander model [10, 11 I, (d) folded chain fringed micellar grains [121,
(e) fringed micellar domain structure (13).
16
of close range order . Their amorphous state is not principally but only gradually
different from a well ordered state. Figure 2.1 shows model representations of the
various interpretations of the nature of the amorphous state:
a) interpenetrating random coils [4], [7]
b) and c) honeycomb and meander m0!iel [10], [11]
d) folded chain fringed micellar grains [12]
e) fringed micellar domain structure [13].
In agreement with the considerations above the term amorphous is used to in-
dicate the absence of crystallographic reflections. If we wish to place importance on
the absence or presence of a close range order within amorphous regions we will
explicitly say so. In the following discussion ordered structural elements are to be
characterized. In doing so one should keep in mind that for the interpretation of
ordered structures in real polymer solids one finds a situation somewhat inverse to
the interpretation of the amorphous state, i.e. one should keep in mind that in any
real polymer well ordered regions are of limited size and finite degree of perfection.
B. Crysta llites
From the beginning of polymer science it had been known that native as well as
synthetic high polymers crystallize [14a]. Analysis of x-ray diffraction patterns has
revealed the lattice structure and unit cell dimensions of high polymer crystallites.
Up to 1957 it was believed that the crystallites are of the fringe micelle type. A
typical micelle was supposed to bundle some tens of hundreds of different molecules
which - after leaving the micelle and passing through amorphous regions - would
at random join other micelles. In 1957 Fischer [15], Keller [16], and Till [17] in-
dependently from each other discovered and proposed that high polymers form single
crystal lamellae within which chain folding occurs. Figure 2.2 shows an electron
micrograph of a stack of PE single crystals [18] grown from dilute solution and
Figure 2.3 the arrangement of the chain molecules in such lamellar crystals. Here
Fig. 2.2. Polyethylene single crystals grown from dilute solution (Courtesy E. W. Fischer, Mainz).
17
Fig. 2.3. Chain folding in polyethylene single crystals.
the PE chain is folded - with adjacent reentry of the chain after each fold - within
the (110) plane of the ortho-rhombic PE crystal. The unit cell dimensions have been
determined [19] to be
a = 0.74 nm
b = 0.493 nm
c = 0.353 nm (chain axis direction).
The largest of the lozenge-shaped single crystals shown has an edge length of
4700 nm and a thickness of about 10 to 15 nm (originally the crystal seems to
have been a hollow pyramid which became flattened during preparation). Lamellar
crystals - including extended chain crystals - of various degrees of perfection and
size are the product of crystallization from solution or from the melt under appro-
priate conditions. The extreme mechanical anisotropy of the chain molecules is
reflected by that of their crystallites. For example the elastic response of PE single
crystals in the three axis directions differs by two orders of magnitude. The elastic
modulus is found [20] to be:
Ea = 5.7 GN/m 2
Eb= 2.1 GN/m 2
Ee = 180-340 GN/m 2
C. Superstructure
At this point the superstructure or morphology of polymers is reviewed with a desire

to identify and define elements of their heterogeneity. The most important hetero-
geneities derive from the tendency of many polymers to (partially) cry&tallize. More
or less well defmed crystalline lamellae are found in form of single crystals as de-
scribed above, stacked and/or grown upon each other in the form ofaxialites or
sheaving layer structures, as twisted aggregates in spherulites, and in the form of
sandwich-like structures in highly oriented fibrils [1-3]. The radially symmetric
growth of twisting lamellae (Fig. 2.4) from many nuclei which leads to the spherulitic
structure, is shown in Figures 2.5. This is the largest and dominant feature of melt-
crystallized specimens.
18
Fig. 2.4. Detail of spheftllite in polyethylene Fig. 2.5. Spherulitic structure of melt crystal-
showing stacks of crystal lamellae (Courtesy lized polyoxymethylene.
E. W. Fischer, Mainz).
Presently it is discussed [91,92,100] whether and to what extent the lamellar

crystals of a solid semicrystalline polymer show the same regularly folded structure
as solution grown single crystals. Schelten et al. [91] concluded that the results of
the small-angle neutron scattering of deuterated PE molecules (PED) in a protonated
PE host (PEH) are incompatible with a regularly folded morphology. Yoon and Flory
[92] supported Scheltens conclusions. They compared the intensity distributions
calculated for several morphological models with the experimental scattering data
[91] . The best fit could be obtained for a model where 30% of "the chains emanating
from a crystalline layer escape therefrom instead of re-entering the layer at the ad-
jacent lattice site". After extended analysis of these and additional own data, Fischer
[100] stated that the radii of gyration (of quenched PE, of PEO crystallized by slow
cooling, and of isothermally crystallized isotactic PP) were only slightly affected
by crystallisation or annealing even though drastic changes of the long spacing
(12-25 nm) had occurred. With regard to the problem of chain folding and adjacent
reentry, the majority of the results seemed to indicate that the adjacent reentry model
is not compatible with neutron scattering behavior. On this basis Fischer [100] ad-
vanced a "solidification-model of crystallization" which excludes large scale diffusion-
al motion of chain segments during crystallization.
The boundaries between spherulites resemble grain boundaries; these grain bound-
ary zones are enriched in low molecular weight material, impurities, chain ends, and
defects. The deformation and strength of such a "composite" structure naturally
depend on the compliances of all of its components. In such a composite large com-
pliances (small constants of elasticity) have to be assigned to the cohesion across
19
grain boundaries and across lamellar fold surfaces. The cohesion between chains
within a crystal lamella is much stronger than the inter crystalline interaction; this
renders a certain stability to the lamellar elements in sample deformation. The de-
formation behavior of such a non-oriented semicrystalline polymer will, therefore,
depend much more on the nature of the secondary forces between the structural
elements than on the length or strength of the chain molecules.
The different forms of superstructure are related to the structure and compo-
sition of the molecular species involved, and to the conditions of polymerization,
processing, and environmental treatment, i.e. on nearly every parameter involved
in producing a polymeric solid. For illustration of the size and shape of supermolec-
ular structures formed during the initial phase of polymerization of a homopolymer
attention is called to the fibrous or globular structures reported by Wristers [21] for
polyolefines. These structures are obtained after polymerization in gaseous or liquid
environment and at low or high efficiency of various Ti-, V-, Cr-, or AI-based cata-
lysts. In Figures 2.6 to 2.8 electron micrographs of such nascent polymer structures
are reproduced [21]. Polypropylene at low catalyst efficiency is formed in globules
of 0.5 J.llIl diameter (Fig. 2.6), at high catalyst efficiency fibers several J1m long are
observed (Fig. 2.7); the fiber diameter is closely correlated with the lateral face
dimension of the primary catalyst crystal and varies between 0.37 and 2 J.llIl for
TiCl 3 -crystal breadths of between 5 and 50 nm. Polyethylene samples prepared with
TiCI 4 -AI(i-Buh or other colloidal catalysts show less regular surface structures
(Fig. 2.8).
Fig. 2.6. Nascent polymer morphology: poly- Fig. 2.7 . Polypropylene prepared at high catalyst
propylene prepared at low catalyst efficiency efficiency (taken from Ref. (211).
(taken from Ref. [21]).
In the course of the bulk polymerization of lactams it was observed [22] that
the resultant polymers precipitated in the form of spherulites. Other examples are
the fibrous, globular, or star-shaped aggregates existing in PVC [23-26] which have
sizes of between 10 and about 500 nm. In good solvents the largest aggregates dis-
integrate into single molecules but only at temperatures larger than 200°C [25].
20
Fig. 2.8. Polypropylene prepared with

TiCl 4 - Al(i-Buh (taken from Ref. (211).
A large group of multi-phase polymeric systems is formed by the block co-

polymers, graft copolymers, random copolymers, and poly-blends. This group of
polymers has received increasing attention (e.g. 27-29) because of the controllable,
"tailor-made" properties of these engineering materials. Here the superstructure is
determined by the relative weight , sequential order, and limited miscibility or even
incompati bility
of the covalently connected molecular segments of hard (glassy, thermoplastic)
and soft (elastomeric, rubbery) chains as within the (two- or three-) block copoly-
mers and random copolymers
of the (two) phases grafted onto eachother within graft-copolymers or
of the molecules mixed together in poly-blends.
The variety of ensuing multi-phase structures cannot be discussed here in any
detail; it ranges from the extremely ordered hexagonal patterns of spherical domains
of one phase embedded within the continuous phase (mostly of the dominant com-
ponent) to the partial phase separation of solution-cast poly-blends occurring only
after heat treatment [27-29]. In Figure 2.9 an electron micrograph of the morpho-
logy of a butadiene styrene diblock copolymer is reproduced [31]. At 17% by weight
the styrene forms the discrete phase of spherical or cylindrical aggregates in a con-
tinuous butadien phase. Owing to the homogeneity of block lengths the aggregates
are arranged in highly regular lattices. The sizes of the aggregates correspond very
well to the molecular conditions. They must be larger than one diameter of the coil
formed by the PS segment and must be smaller than two such diameters to permit
phase separation while maintaining material continuity [28-30].
For a study of the effect of chain length and strength quite naturally structures
which contain a high degree of oriented chains are more promising than others. One
of the most widely investigated ones are the semicrystalline fibers. The characteristic,
microfibrillar and sandwichlike structure of highly oriented fibers will be discussed
in connection with the electron spin resonance investigations of chain scission es-
pecially observed in these fibers.
Still higher degrees of chain orientation are found in the extended chain crys-
tallites, row structures, and highly lamellar "hard-elastic" or "springy" forms of
crystallizable polymers which recently have become of considerable interest. Follow-
21
Fig. 2.9. Thin film of butadien-styrene diblock copolymer cast from petrol ether solution and
annealed 1 hat 100 Cc. At 17% by weight the styrene aggregates in the form of spheres and of
cylinders (Courtesy G. Kampf, Uerdingen).
ing early investigations of Keller [32], Pennings et al. [33] and Kawai et al. [34]
much work has been devoted to the row structures or shish-kebabs which consist
of ribbon-like backbones overgrown by folded chain platelets. Each backbone fiber
again seems to be composed of ultimate fibrils (microfibers) less than 10 nm in
diameter [35]. These structures do form in crystallization under molecular orienta-
tion, in sheared dilute solution as well as in extrusion from the melt [36]. Com-
prehensive reviews on these microfibers have recently been given by Tucker and
George [35-36]. According to them the transverse dimension of these microfibers
is of the order of 0.8 to 20 nm in various celluloses, and between 7 and 50 nm in
most thermoplastic fibers; rnicrofiber lengths are given as 18 to over 1000 nm [35].
At the time of this writing Zwijnenburg and Pennings [37] report on a technique
to obtain PE macrocrystals of virtually infinite length. There, a seed crystal and -
subsequently - the growing crystalline PE fiber is pulled under steady-state con-
ditions from a flowing solution. The diameter of the obtained fiber, an agglomera-
tion of shish-kebab fibrils, varied between 10 and 40 Jilll [37].
In the more recent years the solid state extrusion of thermoplastic polymers
involving very high pressure (up to 5 kbar = 0.5 GN/m 2 ), temperatures between
30 and about 250°C, and extensional flow has been studied [38-44]. In the case
of PE such a treatment led to a new high-pressure, high-temperature phase of hexag-
onal symmetry at a high level of chain extension. In analogy to the extended-chain
crystals observed by e.g. Anderson [45] in the fracture surfaces of low-molecular
weight PE the term extended-chain crystals is presently being used for pressure crys-
tallized PE also. Weeks and Porter found that highly oriented strands of such material
(Mw = 58000) have an exceptional stiffness of 70 GN/m 2 at room temperature
which is comparable to that of E-glass [40] . In addition a good tensile strength of
500 MN/m 2 is reported [41]. Lupton and Regester [39] worked with ultrahigh-
22
I. Elements of the SuperstrUcture of Solid Polymers
molecular-weight (UHMW) HDPE with Mw between 2 and 3· 106 g/mol and ob-
tained a stiffness of 2 to 3 GN/m 2 and a strength of 33-39 MN/m 2 , but an excel-
lent tensile-impact resistance of 300 to 600 kJ/m 2 • Capaccio and Ward [93-98]
prepared ultra oriented polymers (PE, PP, POM) by means of cold drawing and
hydrostatic extrusion. A fibrillar structure seemed to be retained in those samples.
With increasing draw ratio the morphology would be characterized primarily by a
continuum of oriented material in which a statistical distribution of chain ends
would constitute the only discontinuities. Thus an increasingly smaller fraction of
the material would show the conventional morphology comprising crystals and
oriented amorphous chains, including tie molecules, in series.
In the last few years it has been discovered that by controlled drawing and an-
nealing treatment of a number of fiber forming thermoplastics a highly crystalline
morphology with "springy" or hard elastic response could be obtained [46-51,
105]. Springy behavior, the characteristics of which will be discussed in the next
section, has been found so far in PP, POM, PE, PMP, PES, and even in PA 66 [47,
105]. The morphology is proposed to consist of stacks of flat, highly regular,
c-axis oriented folded chain lamellae with the lamellae interconnected at points
about 100 nm apart [46-51,105].
D. Characterization
The experimental methods available to determine or characterize the structure of

polymer chains or of their aggregates are discussed in the general references of
Chapter 1. In addition information on X-ray diffraction [3], neutron scattering
[4-6], electron- and light-scattering [4, 52, 53], optical and electron microscopy
[3, 14b], thermal [3, 54] and viscoelastic behavior [14c, 55-57], and nuclear mag-
netic resonance (NMR) technique [3] can be obtained from the references of this
Chapter. Infrared absorption (IR) and electron spin resonance (ESR) techniques
will be introduced in Chapters 5 and 6 respectively. Except for the microscopic
methods all of the mentioned techniques provide an average information, an in-
formation to which all structural elements within the active volume of an investi-
gated sample contribute according to their individual response, concentration, and
position. Any macroscopic information which is not solely taken as such but is used
to characterize an inhomogeneous solid generally involves elemental response func-
tions and orientation distributions which are not completely known. For this reason
a certain ambiguity is introduced into the deconvolution of such experimental data.
This leaves some room for individual interpretations as to the concentration and
nature of structural elements. This ambiguity has been inherent in the discussion
of the existence or non-existence of close range order in amorphous polymers and
of the nature and distribution of defects within crystallites. Of even more concern
for the structural interpretation of mechanical properties is the fact that the nature
of the connecting members between structural elements is much less well known
than the nature of the latter. This is true on all levels: for the arrangement and inter-
action of non-ordered chain segments, for the fold surfaces of lamellar crystals and
the connections between them, for the amorphous regions between crystallites in
23
fibers or unoriented poly crystalline samples, and for the boundaries of microfibrils
or spherulites. This will be further illustrated in the next section.
II. Deformation
A. Phenomenology
The deformation of the complex structures described in the last section involves a
number of different - and characteristic - phenomena ranging from a linear elastic
to a highly viscous response. Figure 2.1 0 gives examples of the different types of
quasi-static uniaxial-deformation behavior of thermoplastic polymers in the non-
rubbery temperature region. The stress-strain curve a) is that of a brittle polymer
(polystyrene at room temperature) characterized by its very limited extensibility
and the steep and monotonous increase of stress with strain. Curve b) represents a
hard elastic polymer, lamellar polypropylene film [58], which combines strong
elastic behavior with good extensibility at a high stress level and almost complete
recovery (within a period of days). Curve c) is that of a ductile polymer. The initial
monotonous increase of the engineering stress generally is less steep than that of a
brittle polymer, i.e. the secant modulus is smaller. The engineering stress a reaches
a maximum at the yield point (Streckgrenze) which marks the beginning of the so-
called cold drawing, a notable reduction in cross-section of the strained specimen.
If the stress-strain curve extends considerably beyond the yield point one speaks of
a tough polymer; the shape of the curve depends on the strain hardening behavior
of the polymer and on the tendency to neck formation. Curve d) is characteristic
for soft polymers, e.g. for thermoplastics at a temperature close to their glass transi-
tion temperature or plasticised polymers, which do not yet show rubber elastic be-
havior.
As indicated earlier the different stress-strain curves are not characteristic for
particular, chemically defined species of polymers but for the physical state of a
polymeric solid. If the environmental parameters are chosen accordingly transitions
from one type of behavior (e.g. brittle, curve a) to another (e.g. ductile, curve c)
will be observed. These phenomenological aspects of polymer deformation are dis-
cussed in detail in [14], [52-53], [55-57], and in the general references of Chap-
ter 1. A decrease of rate of strain or an increase of temperature generally tend to
increase the ductility and to shift the type of response from that of curve a) towards
that of curves c) and d). At small strains (between zero and about one per cent) the
uniaxial stress a and the strain € are linearly related (Hooke's law):
(2.1)
Even at these small deformations the apparent Young's modulus E is a function

of the rate of strain. This shows that E is not solely determined by the energy elastic
deformation of bond angles, bond lengths, and intermolecular distances but also
involves time-dependent displacements of atoms and small atom-groups. In the fol-
24
II. Deformation
lowing region (1 to about 5% of strain) stress and strain are no longer proportional
to each other, structural and conformational rearrangements occur that are me-
chanically although not thermodynamically reversible, one speaks of an elastic (vis-
coelastic in the narrow sense) or paraelastic behavior. Beyond the yield point large
scale reorientation of chains and lamellar crystals begins, a process generally termed
plastic deformation. A true plastic deformation can be understood as the transfor-
mation from one equilibrium state to another equilibrium state without the creation
of frozen-in tensions. The latter point is particularly important in view of the fact
that the average post-yield deformation is brought about largely by mechanically
reversible anelastic conformational changes of the molecules and not by flow of
molecules past each other. Unless a state of equilibrium had been established through
proper annealing treatment highly drawn samples may retract considerably. Within
the context of this book emphasis should be placed, however, not so much on the
processes leading to or accompanying molecular reorientation - which is basically
a reinforcing effect - but on the damaging processes of chain scission, void forma-
tion, and flow. The latter processes gradually occur in the strain region from just
before the yield point up to final fracture. Among the damaging processes one has
also to include the phenomenon of normal stress yielding or crazing of glassy poly-
mers which will be discussed in Chapter 9.
Not represented in Figure 2.1 0 is the deformation behavior of cross-linked net-
works of highly flexible elastomeric chains. The characteristic tensile deformation
)
of elastomers is not based on energy elasticity but rather on the change of entropy
accompanying the deformation and orientation of randomly coiled chain molecules
60
MN/m2
50
rF PS at200C
c) PVC at 20·C
b) Hard elastic PP/

/
/
/
---
,/
10- ..-- .-/
- --- d) PTFE at 100°C
5 10 15 20 25 'I, 30
strain I £
Fig. 2.10. Characteristic uniaxial deformation behavior of polymers. a. brittle (PS at room-tem-
perature), b. hard elastic (lamellar PP film), c. ductile (PVC at room-temperature), d. soft (PTFE
at 100°C).
25
[59]. To a first approximation stress, uniaxial extension ratio A, and shear modulus
G are related by
(2.2)
The shear modulus G is proportional to the absolute temperature T, Boltzmann

constant k, and the number N of subchains between cross-links:
G = NkT = pRT (1-2 Me/M)/Me (2.3)
where M and Me are the number-average molecular weights of the uncrosslinked

polymer and of the sub chains respectively, p is polymer density.
The pure rubber-elastic deformation is mechanically completely reversible and
does not involve chain scission or creep. A real rubber, however, like any viscoelastic
solid, shows on a molecular scale energy and entropy elastic deformation accompa-
nied by viscous flow. Thus it shows stress relaxation at constant strain, creep at
constant load, and energy dissipation in dynamic excitation. Representation of the
macroscopic mechanical response of viscoelastic solids - even in a strain region
where no large-scale reorientation occurs - requires, therefore, the use of damped
elastic elements consisting of springs (modulus G) and rate sensitive damping ele-
ments (dashpots characterized by a viscosity 1/). The simplest elements are the Max-
well element with spring (G) and dashpot (1/) in series and the Voigt (Kelvin) element
with spring G and dash pot 1/ in parallel. The Maxwell element has a relaxation time
T = 1//G, the Voigt-Kelvin element has the same relaxation time which here more
precisely is called retardation time. The phenomenological theory of viscoelasticity
[55] describes the mechanical response of a body in terms of a distribution of basic
viscoelastic elements primarily characterized by their relaxation times Tj. If the
spectra of the molecular relaxation times, H (1m), are known, then the viscoelastic
modulus functions are, in principle, derivable therefrom [14c, 14d, 55]. The time-
dependent shear or stress relaxation modulus G(t) becomes in terms of a continuous
spectrum of relaxation times:
+=
G(t) = Ge + f H exp(-tIT) dIm (2.4)
where Ge is the modulus at infinite times which - in the case of a solid - is sup-
posedly larger than zero. The modulus functions are conveniently treated as complex
quantities consisting of a ( recoverable) storage and an (irrecoverable) loss compo-
nent; the ratio of loss (e.g. Gil) and storage component (G') is called the loss tangent
or loss factor, i.e. tan 0 = Gil IG'. The phenomenological theory of viscoelasticity is
based on the Boltzmann superposition principle, i.e. on the linear additivity of the
effects of mechanical history. It should be emphasized that the aim of that theory
is not so much to derive the form of the relaxation time spectra from detailed struc-
tural considerations but rather to utilize the response functions obtained in one
experiment, e.g. creep, to predict the response of the same material under different
excitations, e.g. dynamic loading [14, 55]. The general application of this approach
26
II. Deformation
is restricted to the linear range of response. The theory has been extended into the
non-linear range [14d], it has also been adapted to account for limited changes in
the structure and/or of the orientation of the stressed system [l4e]. In the context
of this book use will be made of the established results of the phenomenological
theory as well as of the "molecular theory" of viscoelasticity [55-57], particularly
of the molecular friction coefficient introduced in it.
B. Molecular Description
A kinetic theory of fracture intends to interrelate the motion and response of mole-
cules to the ultimate properties of a stressed sample. A kinetic theory, therefore,
entails a molecular description of the deformation of the microscopically hetero-
geneous and anisotropic aggregates of chains to such an extent that critical defor-
mation processes can be identified. The macroscopic deformation of any aggregate
potentially involves the deformation, displacement, and/or reorientation of so dif-
ferent substructural elements as bond vectors, chain segments or crystal lamellae.
The molecular origin of observed deformation mechanisms have been elucidated
by various spectroscopic methods - including mechanical relaxation spectroscopy -
and by the previously listed tools which characterize the morphology. They are
supplemented by dynamic scattering and diffraction techniques [52, 53].
The following molecular description of the uniaxial straining of unoriented
semicrystalline polyethylene characterizes the ductile deformation of fiber-forming
spherulitic thermoplastics and may serve to illustrate the large variety of deforma-
tion mechanisms involved. At small strains of less than one per cent one observes
the anisotropic elastic response of the (orthorhombiC) PE crystallites [57] and of
the amorphous material [53]. At the same time those anelastic deformations of
CH2 -groups and chain segments occur that characterize the low temperature (3-, 'Y-,
and o-relaxation mechanisms [10,56]. At larger strains (1 to 5 per cent) additional
chain segments change their relative position and conformational changes are ini-
tiated (bond rotation). A detailed study on the behavior of the chains within amor-
phous regions had been carried out by Petraccone et al. [53]. In crystalline regions
subjected to strains of this order, dislocations and dislocation networks are generated
(in lamellar crystals observed by means of moire patterns). Depending on environ-
mental parameters and the type of the dislocations their movement leads to crystal
plastic deformation through twinning or slip or to a phase transformation from the
orthorhombic to a monoclinic unit cell. An extensive review on the deformation
behavior of polymer single crystals has recently been given by Sauer et al. [57] and
in the book of Wunderlich [3]; a detailed account of the contribution of the various
structural elements and defects to the deformation of semicrystalline polymers is
found in the literature in a large number of papers of which only a few have been
cited here [47-62]. Although the above mentioned effects lead to a non-linear stress-
strain response, the initially present substructure is still maintained. Such a defor-
mation is called homogeneous.
Still larger stresses cause a destruction of the substructure, involving reorienta-
tion of chain segments and of lamellar crystals (crystal rotation, chain tilt and slip),
27
opening-up of voids, and the first breakages of chains take place. These processes
account for the ductile deformation. At this stage the mechanical energy input is
largely dissipated as heat as shown in later Chapters. Since the deformation proceeds
at an almost constant level of engineering stress it has generally been termed plastic
deformation - although on a molecular basis it does not correspond to the plastic
deformation of metals. The initial post-yield deformation of PE is predominantly
not brought about by mechanically irreversible flow. In a strain region of up to
about 50 per cent the lamellar crystals slip and/or orient their a-axes perpendicular
to the draw direction. The c-axes (chain axes) resume a preferred angle of 35 to 40°
with respect to the draw direction. The latter orientation, however, is reversible and
changes into a random distribution if external stresses are removed [61]. In a strain
region of between 100 and 400 per cent an increasing c-axis orientation in the draw
direction is observed. In this strain region the polymer undergoes the important
transition from a spherulitic to a fibrillar structure. According to Peterlin [62] micro-
fibrils with lateral dimensions of 20 to 40 nm are formed which contain nearly un-
modified blocks of folded chains broken away from crystal lamellae but still inter-
connected by unfolded sections of tie molecules. A model of the fibrillar structure
of a highly drawn fiber is shown in Fig. 2.l1. The originally random distribution of
u~ U crystall ine region

amorphous region
end of microfibril
a b
polyethylene
6 - po lyamide
Fig. 2.11. Models of the structure of drawn semicrystalline polymers, polyethylene: (a) as
drawn [181, (b) annealed (18) 6-polyamide, as drawn.
chain segments and crystal blocks has become highly oriented. The load carrying
capability of such a structure is - as will be studied in considerable detail - many
times larger than that of the unoriented, spherulitic sample.
The described molecular deformation processes affect the macroscopic fracture
behavior of a given high polymer. The relative magnitude of the deformation com-
pletely determines:
the level of stress and strain to be attained in "homogeneous" deformation
the volume density of elastically stored energy
the rates of energy dissipation and local heating
network reorientation and possible strain hardening
chain scission, void formation, and other forms of structural weakening.
28
III. Model Representation of Deformation
For the interpretation of the complex mechanical response of a highly anisotropic

polymeric network it is indispensable to have a simplifying model representation
of the arrangement and interaction of the structural elements and of their deforma-
tion. Such model representations will be useful in the further investigations which
necessarily focus on some aspects while averaging or neglecting a large number of
others. In this section one would like to characterize the assumptions made on the
geometry and mode of interaction of structural elements in basic theories of the
deformation of a composite body. These theories had been developed to account
for the behavior at small strains. They can be extended into theories of strength if
and only if failure criteria are introduced which become effective within the range
of validity of the particular deformation theory.
There is a large body of literature on the purely elastic interaction of non-pris-
matic structural elements within a deformed "composite" solid (e.g. 63-70); this
literature also forms part of the theoretical foundation of the behavior of viscoelastic
composite solids. While not generally discussing these foundations three groups of
model descriptions of multi-phase materials will be singled out:
those of Voigt [63], Reuss [64], or Takayanagi [71] which assume a homogeneous
distribution of strain or stress,
the calculations of Kerner [65], Hashin [66], van der Poel [67], or Uemura et al.
[72] which minimize the elastic energy of a system containing assumedly spheri-
cal inclusions, and
the empirically modified models of Eilers [68], Guth [69], Mooney [70], and
Nielsen [73] which are based on the original calculations of Einstein [74], where
the disturbances of an elastic field in the vicinity of - and caused by - completely
rigid, spherical inclusions is considered.
It is to be investigated which chain properties are effectively recognized by
these groups of model descriptions of polymer deformation. One knows that the
semicrystalline samples discussed previously are poly crystalline solids containing
dispersed amorphous regions with frequently ill-defined boundaries and equally ill-
defined interaction between amorphous and crystalline regions. In a direct approach
Takayanagi [63] represented semicrystalline polymers by "diagrams" (Fig. 2.12) in-
volving straight boundaries and conditions of plane stress or strain. He obtained a
t stress
4l amorphous
Fig. 2.12. Takayanagi diagram of semicrystalline polymer; con-

Takayanagi dIagram tent of amorphous material corresponds to A . <p.
29
meaningful interpretation of the relative contributions of crystalline and amorphous

phases to the complex moduli of these materials. The Takayanagi approach can well
accomodate information on the deformation behavior of individual crystalline [57]
or amorphous regions [53] that becomes available from other sources. In the analysis
of e.g. the response of a fiber to dynamic loading through a Takayanagi diagram the
nature of chain segment orientation is considered in detail but likely distributions
of chain length and axial chain stresses do not enter the model.
The two -shell model of Kerner [65] conforms to the conditions of the second
group of models. The dilatation of a spherical inclusion surrounded by a homo-
geneous medium is derived subject to the condition that displacements and tractions
at the surface of the inclusion are continuous. The homogeneous medium is supposed
to have the elastic properties of the composite as a whole. The model interrelates
shear (G j ) and compressive (K j ) moduli (or Poisson's ratios Vj) of an arbitrary num-
ber of isotropic elements with the macroscopic moduli G c and Kc.
In the case of isotropic inclusions of an elastic shear modulus Gf comparable to
that of the matrix, G m , Eq. (2.5) describes the variation of the complex shear
modulus, G~, of the composite as a function of the volume content Vf of the dis-
crete phase:
G~= (l-Vr)G~+(a+Vf)Gf
(2.5)
G~ (1 +aVr)G~ +a(l-Vf)Gf
The quantity a is derived from the Poisson's ratio vm as:
(2 .6)
As will be noted no molecular anisotropies and no effects due to size and size
distribution of the particles of the discrete phase are recognized . Through
E* = 2(1 + v)G* Eq. (2.5) can be used to predict also the complex tensile modulus.
A good example for the applicability of Eq. (2.5) is furnished by the experimental
data obtained by Dickie et al. [75] . For the dynamic tensile modulus of a physical
mixture (polymer blend) of 75 % by weight polymethylmethacrylate (PMMA, con-
tinuous phase) and 25% butylacrylate (PBA, discrete phase) within experimental
error correspondence of calculated and measured data was obtained (Fig. 2.13 ,
'"....E
z
(!)
- 0
<0
-1 c:
W !l
.§'
-2 Fig. 2.13. Complex dynamic tensile modulus
of blends (solid line) and graft copolymers
-3 -3 (circles) respectively of polymethyl-methacrylat e
-OJ -SO 0 SO 100 150 'C (75 % by weight) and polybutylacrylate (25 % by
TemperaturE.'
weight) at 110 Hz (after [75, 76)) .
30
solid curves). A copolymer of equal volumetric composition, where 25% by volume

of the elastomeric butylacrylate had been grafted onto PMMA, yielded the experi-
mental data points shown. In the region between the glass transition temperatures
of the PBA (- 20°C) and of the PMMA (at these conditions 160°C) a difference
between the response of the mixture and of the graft copolymer can be observed.
This difference can be related to the existence of the primary bonds which connect
matrix and inclusion and which lead to frozen-in stresses.
A three-shell model of a two-phase solid was proposed by van der Poel ([67],
Fig. 2.14). An interior sphere representing a hard component (G f , Kf , Vf ) is sur-
rounded by a soft shell (G a , Ka , Va), and by a continuum, for which the elastic
Fig. 2.14. Arrangement of phases in the three-

shell model of Van der Poel [671; bulk and
shear moduli (K, G) and Poisson's ratios vof
filler (index f) and matrix (index a) and volume
concentration Vf enter calculated composite
moduli K and G (after 176]).
moduli G and K are being calculated in terms of the mentioned elastic constants
and of the volume concentration (Vf ) of the hard component. The condition of
mechanical equilibrium under shear and hydrostatic pressure leads to an explicit
expression for K and to a complicated determinant for G which has to be solved
numerically and for each individual system. The van der Poel approach had yielded
the best results in describing the elasticity of elastomers highly filled (up to 50 vol.%)
with mineral fillers (see e.g. [77]). Again no orientation distributions or physical an-
isotropies of the constituent particles enter the model.
As mentioned frequently the mechanical and optical response of molecules -
and of their crystallites - is highly anisotropic. Depending on the property under
consideration the carriers of the molecular anisotropy are the bond vectors (infra-
red dichroism), chain segments (optical and mechanical anisotropy), or the end-to-
end vectors of chains (rubber elastic properties). For the representation of the ensu-
ing macroscopic anisotropies one has to recognize, therefore, the molecular aniso-
tropy and the orientation distribution of the anisotropic molecular units (Fig. 1.9.).
Since these are essentially one-dimensional elements their distribution and orienta-
tion behavior can be treated as that of rods; such a model had been used successfully
to explain the optical anisotropy [78], and the anisotropies of thermal conductivity
[79], thermal expansion or linear compressibility [80], and Young's modulus [59,
65,80,81] of rubbers and oriented thermoplastics. The rod models focus on the
large anisotropies which exhibit properties such as the polarizability or the retractive
31
elastic forces; only two components of such a property, namely that in chain axis
direction and that perpendicular to it enter the model as independent parameters.
For example the macroscopic elastic response of an aggregate of randomly,
partly, or well oriented chains can be represented by that of a corresponding dis-
tribution of one-dimensional finite elements (Fig. 2.15) if the following conditions
are fulfilled:
external forces P cause a homogeneous local strain which determines the axial
stress 1/1 ({), I.{J, t) of a chain segment
forces between chains are not transferred by shear or bending moments
the state of stress at any arbitrary point A is given as the integral over all the
traction forces exerted by the chain segments intersecting the surface of a volume
element enclosing the point A [82].
Fig. 2.15. Application of the condition of homogeneoLls strain E (P, t) to a partially oriented
system of finite, one-dimensional elements. The stress tensor ajj is obtained by space averaging
over the traction vectors 'It ({), <p, t).
The components of stress of such a homogeneously strained, partly oriented

aggregate of elastic elements with axial Modulus Eb length L, volume concentration
A, and orientation distribution p was derived by Hsiao [82] as a special case of the
Voigt-model employing a homogeneous distribution of strain:
(2.7)
Here the summation convention of repeated indices is used. The Sj are the com-
ponents of the unit vector in the principal axis directions and w the solid angle. Into
this model only one molecular elastic component enters: the axial chain modulus.
Any interactions by shear or in a direction perpendicular to the chain axes cannot
be accounted for. It should only be applied, therefore, if these interactions can be
meaningfully neglected.
A much more general representation of either different or differently oriented
molecular domains can be achieved, of course, through three-dimensional elements.
In the case of transverse symmetry the molecular elements must be characterized
32
by 5 elastic constants (compliances), the orientation of one or two axes, and the
condition of stress and strain at the boundary of an element. Voigt [63] based his
calculations on the assumption that there be no discontinous change of strain at
any boundary; Reuss [64] made the assumption of homogeneous stress. Following
either Voigt or Reuss the space averaging over the elastic constants Cijmn or elastic
compliances sijmn of the molecular domains leads to the upper or lower bound res-
pectively of the macroscopic moduli [83]. For an affinely deforming aggregate of
such elements Ward [84] and later Kausch [85] have calculated the macroscopic
elastic moduli as a function of domain orientation. The calculated curves for the
change of the elastic moduli with draw ratio are particularly characterized by the
rate of change and by the extent of change ultimately attained. If only reorienta-
tion of otherwise unchanged molecular domains occurs in drawing, the properties
of a "completely" oriented sample must correspond to the properties of the mole-
cular domains. In Figure 2.16 the calculated changes of Young's modulus in the direc-
----- "----
)
s"/s,, =
---- -- 1.0
1.0
....~PMMA
~.,-
---- f---- 1--- 1-' 0.5
--
1----
0.8
/~v,s.--- + ----r--[2--
~,/' f-PA";-
-
1----
--- -
// ." ....... ....-+
-- --
,," 1-1i.07"-
0.6
-
/"
",'"
0.4
/+
'" ........ ~ - - - theoretical curves
Nylon 66
,,/ .>- homogeneous stress
Q2
.,.)"""
r-'
o
1 3.5 6 8.5 11 13.5 16 a.
draw ratio
Fig. 2.16. Change of Young's modulus E33 = a33/e33 as a function of draw ratio Qand of the
mechanical anisotropy 533/511 = S3333/s1111 of the molecular domains (after (13), 185]).
tion of draw as a function of draw ratio and domain anisotropy are represented in
comparison to experimental data [13, 85]. The findings of Ward and Kausch may
be summarized by saying:
the orientation behavior of amorphous thermoplastics (PS, PMMA, PVC) could
be represented by the affine deformation of oriented domains
the observed data for PS, PMMA, and PVC were in agreement with the assump-
tion of homogeneous stress at the boundaries of the orienting domains, the data
for PC were better in agreement with a condition of homogeneous strain
the orienting molecular domains have comparatively small elastic anisotropies
(S3333/S1111 -values ranging from 0.5 to 1), the axial compliances being more
than an order of magnitude larger than those of the chains; this indicates that
33
the domains are not equivalent to chains, micro-crystallites, or fringe micelles;
if such domains do have a physical significance in amorphous polymers then they
must be considered to be aggregates of chain segments with a stiffness about twice
as large as that of the sample as a whole [13]
- the concept of a/finely orienting lamellae in semicrystalline polymers is only a
crude approximation [13, 87-89, 106].
The mathematical representation of the elastic behavior of oriented hetero-
geneous solids can be somewhat improved through a more appropriate choice of the
boundary conditions such as proposed by Hashin and Shtrikman [66] and Stern-
stein and Lederle [86]. In the case of lamellar polymers the formalisms developed
for reinforced materials are quite useful [87-88]. An extensive review on the experi-
mental characterization of the anisotropic and non-linear viscoelastic behavior of
solid polymers and of their model interpretation had been given by Hadley and Ward
[89]. New descriptions of polymer structure and deformation derive from the con-
cept of paracrystalline domains particularly proposed by Hosemann [9, 90] and
Bonart [90], from a thermodynamic treatment of defect concentrations in bundles
of chains according to the kink and meander model of Pechhold [10-11], and from
the continuum mechanical analysis developed by Anthony and Kroner [14g, 99].
The theory of paracrystals recognized that chain molecules form three-dimen-
sionallattices and establish a certain long range order between members of the same
lattice while postulating at the same time that in a real solid the lattice will be dis-
torted. The degree of lattice distortion is measured by the variation in length that
the three distance vectors ai between corresponding lattice points encounter if moved
in the three lattice directions. If the dimensionless relative mean fluctuations gik of
the distance vectors 3j are all zero the structure is crystalline and if all gik > 0.1 it
is amorphous. The gik provide a quantitative measure of the colloidal structure of
micro heterogeneous solids. If for instance g13 and g23 are large compared with the
other gik we deal with the nematic state (segments parallel but at random distances),
if g31 and g32 are large compared with the other gik we have a smectic state where
the segments arrange in layers [9]. The relative paracrystalline distance fluctuation
was shown to be inversely related to the maximum number of network planes within
one microdomain [9]. The gik are obtained from line profile measurements in small
angle X-ray scattering (SAXS).
Hosemann's schematic representation of a two-dimensional paracrystalline lattice
is given in Figure 2.17. If we interpret a superstructure (e.g. the fibrillar structure
shown in Figure 2.11) in terms of paracrystals rather than crystals we arrive at the
very same size and orientation distribution of the scattering elements - be they
crystallites or paracrystals. The differences in interpretation become apparent, how-
ever, if intercrystalline rearrangements in annealing, during deformation, or after
irradiation are discussed [9].
The general concept that there are only gradual differences between the more
or less ordered regions of polymers is also employed by Pechhold and his co-workers
in their quantitative microstructural theory (Kinkenmodell) of deformation [10,
11, 14f]. Pechhold recognizes that chain molecules exist in energetically different
states of rotational isomers between which - cooperative and mostly rapid - tran-
sitions can be accomplished. He bases his quantitative thermodynamical calculations
34
1..1
.' "\:1' ....
U A . omorphouS .. pha~ ··
..1 C.F. clu!ltrrd librlll(hO/strl'lctlt'd)
e.G. crystol growth 1ft bulk mol,riol
E. rnd of a chOn
FP. four-point·dlagrom
LA long backfol dl ng {Floryl
I4t m'gratin g lold
P.' paroc~slallin' Ioyu/otlicr
S. slra' I chains
sa 51"" backlolding (K./I.,I
~ ~Wdt: ~;lJi~~~~Old slrrlchf>d)
SH $h~ring rrgion
s.r Sial/on mod.,
V. . olds"
b
Fig. 2.17. Schematic para crystalline (a) and molecular representation of different two-dimen-
sionallattices and superstru ctur es (after (91). a1 = crystal , a2 = ideal paracry sta l, a3 = real para-
crystal, a4 = amorphous state, as = micro-paracrystallites (function of "micellar cross-links") ,
b = molecular model of linear polyethylene.
on kink isomers, which show only a small departure from the extended planar chain
and are most likely to form bundles ([ 11], Fig. 2.18 a). The neighbouring chains with-
in such bundle form various pairs of segments (cf. Fig. 1.12) which have different
volumes, axial lengths and interaction energies. The concentration of kinks deter-
mines the axial extension of the bundle. In the partition function of a bundle of
chains all these quantities enter; the partition function relates, therefore, the kink
35
area
.......- - - - - -
a b
Fig. 2.18. Kink-model representation of (a) bundles of chains and (b) oriented fibers with sand-
wich-like structure (after [10, 11 D.
concentration with bundle geometry and internal and free energy of the isomers.
Consequently the elastic compliances of a transversely isotropic bundle of chains
could be determined as a function of kink concentration, kink block size, width of
kink steps, and temperature [10,11, 14f]. The anelastic and plastic deformation
may be interpreted in terms of successive dislocation motion and slip of bundle sur-
face areas. The kink model offers in every respect a molecular description of poly-
mer structure. Again there are only gradual differences between ordered and non-
ordered regions. Pechhold estimates that an apparently perfect PE-crystallite may
contain up to 4 kinks per 1000 CH 2 -groups whereas in a melt-like structure this con-
centration is about 200 in 1000. Although this concentration is so large as to an-
nihilate all close and long range order some logical principles should govern the space
filling through chain molecules. Pechhold generated appropriate patterns, the honey-
comb- and the meander model (Fig. 2.1 c). He feels that the latter structure is the
more probable one and may exist in semi-crystalline fibers (Fig. 2.18 b) as well as
in rubbers [11, 14f]. The previously mentioned a, (3 and 'Y relaxations are interpreted
in terms of the above model as the motion of kink-blocks, as freezing-in of segment
rotation because of lack of free volume, and as the steps and jumps of kinks in the
amorphous and crystalline regions respectively lll]. Although meanwhile the neu-
tron scattering experiments [100-104] strongly disprove the existence of a distinct
meander arrangement of chains, Pechhold's considerations have greatly fertilized the
efforts to study the structure of amorphous regions.
The close geometrical description of the arrangement of chain molecules within
bundles offered by the kink-model have been used by Kroner and Anthony [14g,
99] to develop a quantitative non-linear deformation theory based on the structure
defect "disclination". The meander model, then, is reduced to a particular arrange-
ment of disclinations.
36
With regard to the elucidation of phenomena occurring during crack propaga-

tion in polymers it is generally not necessary to consider intracrystalline processes in
detail. If, however, the intracrystalline segments of tie molecules are concerned then
one has to be aware of the fact that the intermolecular interaction determines what
forces can be transmitted onto these chain segments and whether or not these forces
are large enough to cause chain scission.
Thus far model representations of polymer deformation have been discussed.
Each of those could be converted into a model representing the fracture behavior
if it were possible to formulate an adequate fracture criterion within the range of
validity of these models. Having dealt with deformation the fracture criteria to be
formulated would have to involve finite extensibility, critical load, or limited volume
concentration of stored or dissipated energy. Dealing with fracture one will find
that strain, stress, and energy are not sufficient as variables and that one will have
to add at least two new dimensions: time and structural discontinuity. This will be
explained in the following Chapter.

1. P. H. Geil: Polymer Single Crystals, New York: Interscience 1963.
2. A. Keller: Polymer crystals, Rep. Progr. Phys. 31,623-704 (1968).
3. B. Wunderlich' Macromolecular Physics: Crystal Structure, Morphology, Defects (Vol. 1),
Crystal Nucleation, Growth, Annealing (Vol. 2), New York and London: Academic Press
1973.
4. E. W. Fischer: Structure of amorphous organic polymers in bulk, Proc. Conf. Non-Crys-
talline Solids, Clausthal-Zellerfeld, Sept. 1976.
5. R. G. Kirste, W. A. Kruse, J. Schelten: Makromol. Chern. 162, 299 (1972).
6. J. P. Cotton, D. Decker, H. Benoit, B. Farnoux, J. Higgins, G. Jannink, R. Ober, C. Picot,
J. des Cloizeaux: Macromolecules 7, 863 (1974).
7. P. J. Flory.: Statistical Mechanics of Chain Molecules, New York: Wiley 1969.
8. P. J. Flory: Spatial configuration of macromolecular chains, The 1974 Nobel Lecture, Brit.
Polymer J. 8,1-10 (1976).
9. R. Hosemann: Makromol. Chern., Suppl.1, 559-577 (1975). R. Hosemann: Ber. Bunsen-
ges. phys. Chern. 74,755-767 (1970).
10. W. Pechhold: Kolloid-Z. Z. Polymere 228, 1 (1968).
11. W. Pechhold: J. Polymer Sci. C32, 123-148 (1971).
12. G. S. Y. Yeh: Polymer Prepr. 14/2, 718 (1973).
13. H. H. Kausch: KollOld-Z. Z. Polymere 234, 1148-1149 (1969),237, 251-266 (1970) and
J. Polymer Sci. C32, 1-44 (1971).
14. a) H. F. Mark: p. xiii. b) H. Glelter, R. Hornbogen, J. Petermann: p. 149. c) J. D. Ferry:
p. 27. d) R. S. Rivlin: p. 71. e) R. F. Landel, R. F. Fedors: p. 131. f) W. Pechhold: p. 301.
g) K. H. Anthony, E. Kroner: p. 429. Deformation and Fracture of Polymers: H. H. Kausch,
J. A. Hassel, R. I. Jaffee, eds., New York/London: Plenum Press 1973.
15. E. W. Fischer: Z. Naturforsch.12a, 753 (1957).
16. A. Keller: Phil. Mag. 2, 1171 (1957).
17. P. H. Till: J. Polymer Sci. 24,301 (1957).
18. Courtesy of E. W. Fischer: Mainz 1974.
19. C. W. Bunn: Trans. Faraday Soc. 35, 482 (1939).
20. T. Miyazawa, T. Kitagawa: J. Polymer Sci. B2 (1964).
21. J. Wristers: J. Polymer Sci., Polym. Physics Ed. 11, 1601-1617 (1973).
22. T. Komoto, M. Iguchi, H. Kanetsuna, T. Kawai: Makromol. Chern. 135, 145 (1975).
23. K. C. Tsou, H. P. Geil: Int.l. Polym. Mater. 1, 223 (1972).
37
24. T. Hattori, K. Tanaka, M. Matsuo: Polym. Eng. Sci. 12, 199 (1972).
25. A. H. Abdel-Alim, A. E. Hamielec: 1. Appl. Polym. Sci. 17,3033-3047 (1973).
26. A. H. Abdel-Alim: J. Appl. Polym. Sci. 19, 2179-2185 (1975).
27. N. A. J. Platzer: Copolymers, Polyblends, and Composites, Adv. Chern. 142 (1975).
28. J. A. Manson, L. H. Sperling: Polymer Blends and Composites, New York and London:
Plenum Press 1976.
29. Mehrphasensysteme, Spring Meeting of the German Chemical and Physical Societies,
Bad Nauheim, March 29 - April 2, 1976, papers published in Angew. Makromol. Chern.
58/59 and 60/61 (1977).
30. G. Kiimpf, M. Hoffmann, H. Kromer: Ber. Bunsen-Ges. 74, 851 (1970).
31. Courtesy to G. Kampf: Uerdingen,1976.
32. A. Keller: J. Polymer Sci. 15,32-49 (1955).
33. A. J. Pennings, A. M. Kiel: Kolloid-Z. Z. Polymere205, 160-162 (1965).
34. T. Kawai, T. Matsumoto, M. Kato, H. Maeda: Kolloid-Z. Z. Polymere 222,1-10 (1968).
35. P. Tucker, W. George: Polym. Eng. Sci. 12, 364-377 (1972).
36. P. Tucker, W. George: Text. Res. J. 44,56-70 (1974).
37. A. Zwijnenburg, A. J. Pennings: Ko110 id-Z. Z. Polymere 254, 868-881 (1976).
38. M. Takayanagi, In: Deformation and Fracture of High Polymers, H. H. Kausch, 1. A. Hassel,
R. I. Jaffee, eds., New York: Plenum Press 1973, p. 353-376.
39. J. M. Lupton, J. W. Regester: J. Appl. Polymer Sci. 18, 2407-2425 (1974).
40. N. E. Weeks, R. S. Porter: J. Polymer Sci., Polymer Physics Ed. 12, 635-643 (1974).
41. N. J. Capiati, R. S. Porter: ibid. 13, 1177-1186 (1975).
42. R. B. Morris, D. C. Bassett: ibid. 13, 1501-1509 (1975).
43. N. E. Weeks, R. S. Porter: ibid. 13, 2031-2048 (1975).
44. N. E. Weeks, R. S. Porter: ibid. 13, 2049 -2065 (1975).
45. F. R. Anderson: J. Polymer Sci. e3, 123 (1963).
46. E. S. Clark, C. A. Garber: Int. J. Polym. Mater. 1, 31 (1971).
47. W. O. Statton: Report UTEC MSE 73-100, University of Utah, Salt Lake City, 1973.
48. H. D. Noether, W. Whitney: Kolloid-Z. Z. Polymere 251, 991 (1973).
49. I. K. Park, H. D. Noether: Colloid and Polymer Sci. 253,824-839 (1975).
50. D. Goritz: Colloid and Polymer Sci. 253,844-851 (1975).
51. B. Cayrol, J. Petermann: J. Polymer Sci., Polymer Phys. Ed. 12, 2169-2172 (1974).
52. R. S. Stein: J. Polymer Sci. e32, 45-68 (1971).
53. V. Petraccone, I. C. Sanchez, R. S. Stein: J. Polymer Sci.,.Polymer Phys. Ed. 13,
1991-2029 (1975).
54. B. Wunderlich, H. Baur: Adv. Polymer Sci. 7, 151- 368 (1970).
55. J. D. Ferry: Viscoelastic Properties of Polymers, New York: Wiley 1970.
56. J. A. Sauer, A. E. Woodward: Polymer Thermal Analysis, II. P. E. Slade, Jr., L. T. Jenkins,
eds., New York: Dekker, Inc. 1970, p. 107.
57. J. A. Sauer, G. C. Richardson, D. R. Morrow: J. Macromol. Sci. - Revs. Macromol. Chern.
e9(2), 149-267 (1973).
58. H. D. Noether: personal communication.
59. L. R. G. Treloar: The Physics of Rubber Elasticity, London: Oxford Univ. Press 1958.
60. M. Maeda, S. Hibi, F. Itoh, S. Nomura, T. Kawaguchi, H. Kawai: J. Polymer Sci., A-2, 8,
1303-1322 (1970).
S. Nomura, A. Asanuma. S. Suehiro, H. Kawai: ibid. 9,1991-2007 (1971).
Rheo-Optical Studies of High Polymers, Onogi Laboratory, Kyoto University, Kyoto,
Japan, 1971.
61. N. Kasai, M. Kakudo: J. Polymer Sci. A-2, 1955 (1964).
62. A. Peterlin: Adv. Polymer Science and Engineering, K. D. Pae, D. R. Morrow, and Yo Chen,
eds., New York: Plenum Press 1972, p. 1.
63. W. Voigt: Lehrbuch der Kristallphysik, B. G. Teubner Leipzig (1910).
64. A. Reuss: Z. angew. Math. Mech. 9, 49 (1929).
65. E. H. Kerner: Proc. Phys. Soc. B 69, 802(956), Proc. Phys. Soc. B 69, 808 (1956).
66. Z. Hashin: J. Mech. Phys. Solids 10, 335 (1962). Z. Hashin, S. Shtrikman: J. Mech. Phys.
Solids 10, 343 (1962).
38
67. C. van der Poel: Rheo!' Acta 1,198 (1958).

68. H. Eilers: Kolloid-Z. 97, 313 (1941).
69. E. Guth: J. App!. Phys. 16, 20 (1954). E. Guth, O. Gold: Phys. Rev. 53, 322 (1938).
70. M. Mooney: J. Colloid Sci. 6,162-170 (1951).
71. M. Takayanagi, K. Imada, T. Kajiyama: J. Polymer Sci., C15, 263-281 (1966).
72. S. Uemura, M. Takayanagi: J. App!. Polymer Sci. 10, 113 (1966).
73. L. E. Nielsen: Mechanical Properties of Polymers and Composites, New York: Marcel
Dekker, 1974, Vol. 2, p. 387 ff.
74. A. Einstein: Annalen Phys.I9, 289 (1906), Annalen Phys. 34,591 (1911).
75. R. A. Dickie, M. Cheung: J. App!. Polymer Sci. 17, 79-94 (1973).
76. H. H. Kausch: Angew. Makromo!. Chern. 60/61, 139 (1977).
77. F. R. Schwarzl, H. W. Bree, C. J. Nederveen: Proc. IV. Int. Congr. Rheology (Providence
1963). E. H. Lee, ed., New York: Interscience, 1965, Vol. 3, p. 241-263.
78. W. Kuhn: Kolloid-Z. 87,3 (1939). W. Kuhn, F. Grun: Kolloid-Z. 1, 248 (1942).
79. K. Eiermann: Kunststoffe 51, 512 (1961). K. Eiermann, K. H. Hellwege: J. Polymer Sci.
1, 99 (1962).
80. J. Hennig: Kunststoffe 57,385 (1967). J. Hennig: Kolloid-Z. Z. Polymere 202,127 (1965).
81. W. Hellmuth, H. G. Kilian, F. H. Muller: Kolloid-Z. Z. Polymere 218,10 (1967). F. H.
Muller: J. Polymer Sci. C20, 61 (1967).
82. C. C. Hsiao: J. Polymer Sci. 44, 71 (1960).
83. J. Bishop, R. Hill: Phil. Mag. 42,414 and 1298 (1951).
84. l. M. Ward: Proc. Phys. Soc. 80, 11 76 (1962).
85. H. H. Kausch: J. App!. Phys. 38, 4213 (1967).
86. S. S. Sternstein, G. M. Lederle, In: Polymer Networks: Structure and Mechanical Properties,
Newman and Chompff, eds., New York: Plenum Press, 1971.
87. J. C. Halpin, J. C. Kardos: J. App!. Phys. 43,2235 (1972).
88. E. H. Andrews: Pure App!. Chern. 39, 179-194 (1974).
89. D. W. Hadley, l. M. Ward: Rep. Progr. Physics, 38,1143-1215 (1975).
90. R. Bonart, R. Hosemann: Kolloid-Z. Z. Polymere 186, 16 (1962).
91. J. Schelten, D. G. H. Ballard, G. D. Wignall, G. W. Longman, W. Schmatz: Polymer 17,
751 (1976).
92. D. Y. Yoon, P. J. Flory: Polymer 18,509-513 (1976).
93. G. Capaccio, I. M. Ward: Polymer 16, 239 (1975).
94. G. Capaccio, T. A. Crompton, I. M. Ward: J. Polymer Sci. A214, 1641 (1976).
95. G. Capaccio, T. J. Chapman, l. M. Ward: Polymer 16, 469 (1975).
96. G. Capaccio, T. A. Crompton, I. M. Ward: Polymer 17,645 (1976).
97. I. M. Ward: Seminar on Ultra-High Polymers, St. Margherita Ligure, May 1977.
98. G. Capaccio: Seminar on Ultra-High Polymers, St. Margherita Ligure, May 1977.
99. K. Anthony: Habilitationsschrift, Universitat Stuttgart, November 1974.
100. E. W. Fischer: Europhysics Conf. Abstracts 2 E, 71-79 (1977).
101. J. Schelten, G. D. Wignall, D. G. H. Ballard, G. W. Longmann: Polymer 18/11,1111 (1977).
102. H. Benoit, D. Decker, R. Duplessix, C. Picot, P. Rempp, J. P. Cotton, B. Farnoux, G.
Jannink, R. Ober: J. Polym. Sci., Polym. Phys. Ed. 14, 2119-2128 (1976).
103. C. Picot, R. Duplessix, D. Decker, H. Benoit, F. Boue, J. P. Cotton, M. Daoud, B. Farnoux,
G. Jannink, M. Nierlich, A. J. deVries, P. Pincus: Macromolecules 10,436-442 (1977).
104. Europhysics Conf. Abstracts 2F (1978).
105. S. L. Cannon, G. B. McKenna, W. O. Statton: J. Polymer Sci., Macromol. Rev.l1, 209-275
(1976).
106. H. G. Kilian, M. Pietralla (Polymer 19, 664-672, 1978) derive from the anisotropy A of
the thermal diffusivity '" = A/Cpp of oriented polyethylenes that the intrinsic anisotropy
Ai of the orienting, partly crystalline lamellar clusters increases with the degree of crystal-
linity (Ai = 7 to 26, linear extrapolation yields Ai = 2 for the fully amorphous and Ai = 50
for a completely crystalline cluster); the average degree of orientation of the lamellae,
< cos 2 6 >, as determined from thermal measurements agrees very well with X-ray data;
the observed increase of orientation with draw ratio is more rapid than it would be in
affine deformation.
39
Chapter 3
Statistical, Continuum Mechanical,

and Rate Process Theories of Fracture
I. Introduction 40
II. Statistical Aspects 41
III. Continuum and Fracture Mechanics Approach . 47
A. Classical Failure Criteria . 47
B. Fracture Mechanics 50
C. Continuous Viscoelastic Models 52
IV. Rate Process Theories of Fracture 53
A. Overview 53
B. Eyring's Theory of Flow 54
C. Tobolsky - Eyring 55
D. Zhurkov, Bueche 56
E. Hsiao-Kausch 59
F. Gotlib, Dobrodumov et al. 64
G. Bueche - Halpin 65
I. Introduction
Test specimens loaded under laboratory conditions, but also regular engineering com-
ponents fracturing in service, provide many different data which lend themselves to
evaluation of the fracture process. These data are, for instance, the time to onset
and completion of fracture, details of the fracture pattern (ductile or brittle frac-
ture, appearance of the breaking specimen and of the fracture surface), crack dynam-
ics, and change in physical or chemical properties. Naturally the most straight-
forward evaluation of a test or a set of data is the direct correlation of the property
of interest (e.g., time under stress) to the environmental parameter(s) of interest
(e.g., stress and temperature). In Figure 1.4 a set of data with just these variables has
been plotted (PVC pipes under internal pressure). If one uses such a plot a number
of questions arise:
what is the statistical significance of an individual data point
what are the failure conditions and how do they depend on external parameters
and material properties
what is the likely cause of failure
what conclusions can be drawn with respect to extrapolating the set of curves
into hitherto non-accessible regions of time, pressure, or temperature?
40
II. Statistical Aspects
The first of these points will be commented upon subsequently in connection

with an interpretation of the statistical variability of fracture events. To answer
this and the second question the highly developed tools of statistical analysis (e.g.,
1-2) are available; their application forms a backbone of industrial quality control
and material design. Understandably, the complex input (mechanical, thermal, and
environmental attack) acting upon a complex system (e. g., a highly structured
polymer) has a complex, non-deterministic output. The determination and evaluation
of only a small number of different data will necessarily reveal only a partial aspect
of the fracture process. For this reason the questions as to cause and kinetics of
fracture development can rarely be answered unambiguously. Any mathematical
interrelations established between variables (e.g., time and stress) are valid for ex-
trapolation only if the basis does not change.
In this chapter an overview of conceptually different fracture theories is pre-
sented which have in common that they do not make explicite reference to the
characteristic properties of the molecular chains, their configurational and super-
molecular order and their thermal and mechanical interaction. This will be seen
to apply to the classical failure criteria and general continuum mechanical models.
Rate process fracture theories take into consideration the viscoelastic behavior of
polymeric materials but do not derive their fracture criteria from detailed mor-
phological analysis. These basic theories are invaluable, however, to elucidate
statistical, non-morphological, or continuum mechanical aspects of the fracture
process.
Experimentally determined quantities (observables) such as strength, time to frac-

ture, or concentration of free radicals show a wide scatter of values; they are sto-
chastic variables. As an extreme example of a stochastic variable in Figure 3.1 the
service lives t of 500 HDPE-tubes under identical test conditions are plotted in the
form of a histogram [3]. The distribution shown can be represented by a logarithmic
normal distribution (Fig. 3.2) with a mean oflg t/h equal to 2.3937 and a variance
of s = 0.3043. The expectation value of the time to fracture of a specimen subject
to the test condition employed is the time which corresponds to the mean value of
the (logarithmic) distribution, here 247.6 h. The actually determined values of t
evidently scatter widely around this expectation value. Nevertheless even such a
distribution can be characterized by testing just a few randomly selected specimens.
For a normally distributed observable any three random values cover on the average
a range of 1.69 s - which contains in 86% of all cases the mean value of the (in-
finite) distribution also [3]. This extreme example has been chosen to emphasize the
meaning of stochastic variables with which one is dealing subsequently. It was also
intended to show the Significance of small samples - containing 3 to 10 members
- which are generally used in practice to establish correlations between stochastic
variables and environmental or material parameters. At this point we are not going
to talk further about the statistical reduction of such a set of data. Here we are in-
terested in the information which may be drawn from such a statistical analysis.
41
3. Statistical, Continuum Mechanical, and Rate Process Theories of Fracture
5;~==~~====~====~,======I======C==
all PE tubes under equal,
N constant Internal pressure
~2 o·
1'":
f
~
." 15
<II
"::>
~
.3 10
'0 I
Ii;
.Q I
E
::> 5 i
co
on
o 200 400 600 800 tb 1000 h
time to fracture
Fig. 3.1. Scatter of times to fracture of 500 PE tubes under identical loading conditions. 80°C,
DV= 4 MN/m 2 (after [3), from [781).
..
c
0
:l
.J:l 95
...
L..
III 80
'" 50
-'
0
~
en 20
CII
.....
c 5
0,8
1.8
logarithm of Lifetime
Fig. 3.2. Cumulative normal distribution of the logarithms of the lifetimes plotted in Fig. 3.1.
From a statistical standpoint, the variability of a material property, e.g., frac-

ture strength, is explained by referring to one or more of the following three ar-
guments:
1. The beginning of fracture is a statistical event, the occurrence of which is de-
scribed by probability laws.
2. Apparently identical specimens of a set are inherently different, e.g., they contain a
large number of flaws of different size or severeness and the most severe of these
flaws determines the fracture strength.
3. The load to which a sample is subjected until it fractures activates a large
number of molecular processes. The uncertainty of the cumulative event, e.g., mac-
42
roscopic failure, then results from the uncertainty of the molceular events and the
mode of correlation of these events.
The first argument (fracture is a statistical event) would lead to the following
general equation for the rate of change of the number N of members belonging to
one ensemble:
dN/dt=-NK (3.1)
where K may be a function of N, stress, stress history, temperature, or other en-

vironmental parameters. For a number N of statistically equal, independent and
separately loaded units the rate function Kin Eq. (3.1) is independent of N, but
generally a function of time. Equation (3.1) is readily solved with respect to the
number N(t) of surviving units:
N(t)::: Noexp [ - ~ K(r) dr] (3.2)
where No is their initial number. The fraction of units broken after some time tb
is obviously 1 - N(tb)/No and the cumulative distribution function of lifetimes,
Q(tb ), therefore is:
Q(t b) =1 - exp [ -? K(r) dr ] . (3.3)
The average lifetime, < tb >, is solely determined by K and becomes for constant K:
< tb > = NI ; tNoK exp (-Kt) dt = K- 1 . (3.4)

o 0
It may be noted that in this case the life expectancy of a surviving unit is at any
time independent of Nand t. This is equivalent to the assumption that the residual
strength is equal to the initial strength, i.e., that no aging or deterioration has taken
place in the stressed sample. The rate K of failure occurrence can be determined
experimentally from the scatter of tb [4]. Using this assumption Kawabata and
Blatz [5] developed a simple stochastic theory of creep failure. They argued that the
first moments of tb for an ensemble of ruptured specimens should reveal whether
or not K is independent of time. If K depends only on initial load and is independent
of N the equation
K::: (m!/< t: »l/m (3.5)
should hold for any m, that is for any order of moments of tb' An investigation of
the variability of fracture times in the special case of a rubber vulcanizate gave ex-
cellent agreement with the assumption of a time-independent K as used in Eq. (3.4).
43
The same conclusion was drawn by Narisawa et al. [82] from the exponential decrease
ofN(t) with t of uniaxially stressed PA 6 monofIlaments.
In other experiments Kawabata et al. [6] investigated the failure statistics of
carbon-filled styrene butadien rubber (SBR). They arrived at the conclusion that
either the coefficient relating stress and rate of failure is increasing with time or
that several local fracture events must be independently initiated before the rubber
specimen breaks. The best fit of their theory with experiments was obtained for a
critical number of 3 to 4 microscopic fracture events as nucleus for unstable crack
formation. For the asymmetric distribution of times-to-fracture shown in Figure 3.2
the relation Eq. (3.5) is also not valid for higher orders of m (m equal to or larger
than two). This means either that the probability density K for pipe failure is smaller
for specimens having a longer service life or that K is a function of loading time. In
the first case it must be assumed that from the beginning the specimens had not
been statistically identical, in the second that they underwent structural changes
affecting K. The first assumption seems to be valid for pipe failure through creep
crack development. The investigations of Stockmayer [83] recently have shown that
nature and location of material inhomogeneities largely determine the creep lives
tb of LDPE pipes. Morphological parameters thus account for more than half of
< t% > [84]. The conclusion that structural differences exist and affect K must be
drawn for other materials as well which exhibit a log-normal distribution of tb values
[3, 83-85]. These aspects will have to be discussed in later sections in terms of
detailed structural models of the failing specimens.
Without reference to any such models and based purely on statistical considera-
tions, Coleman and Marquardt developed their important theory of breaking kinetics
of fibers (reviewed in ref. 7). They especially investigated the distribution of life-
times of fibers under constant or periodic load and the effects of fiber length, rate
of loading, and bundle size on the strength of fIlaments or bundles thereof (Figs.
3.3 and 3.4). Two statistical effects should be noted: the lower strength of a bundle
as compared with that of a monofIlament (due to the accelerating increase in failure
probability K after rupture of the first fIlament within a bundle) and the increase of
strength with loading rate here derived from the reduced time each fIlament spends
at each load level. The authors [8] determined the average tensile strengths of
15-denier 66 nylon monofIls and that of an infinite bundle; the strength values are
plotted as a function of loading rate in Figure 3.3.
As mentioned above the large scatter of strengths and the dependency on sample
geometry may be explained by introducing the concept of flaws of different degrees
of severity. This means substituting one statistical variable (e.g., underlying strength)
for another one (e.g., macroscopic strength or time to failure). Even though such a
substitution leaves the nature of the microscopic defects out of consideration, it
still allows one to obtain, from statistical analysis, some information on the size,
number, location, and severity of these defects (or flaws). The term flaw needs some
deliberation. First, a flaw can be considered as an ellipsoidaUy shaped void which
may act as a stress concentrator and as the eventual cause of instability and failure.
Secondly, it should also be understood as a weak region containing molecular irregu-
larities. When being stressed the weak region may fail, leading to a void, a craze,
and/or to a propagating crack and macroscopic failure. Following Epstein's statistical
44
0*
I I
l-L- --
I lonofiloment
as
0.6
---- --::: 0 ..... \ .....
.0........
1nfinite bundle
I
!
"time"
(l I
-1 3 log a
rate of loading [MN m-2 S-l J
Fig. 3.3. Strength of 66 polyamide mono filaments and bundles of filaments as a function of
loading rate (after Coleman et aI., (81).
0 4 8 12 log (Ln.)
i
12 0
~,x,
0+ l:J.o*
theory ----: -V-
1.0
-2
...... x .. x~
· ..... x ..... .c
"x--... C,
08 c:
!!!
~
Wi
,........,
E
....
Z
06 -4
.
'0
III
tl
.E... .~
u
.c
C,
c: 04 ..
"t>
~
!!! ;;
Wi monofIlament - -6
bundle of 140 monofils ~
0.2
_ _ experImental data
OL-______~+,
o
____~I ________
4
----------~----~--~-8
8 log L(cm) 12
Fig. 3.4. Effect of specimen length Lon 6-polyamide fiber strength (after [11 I); theoretical
curve from Eq. 3.6, experimental data from [101.
theory of extreme values [9] the strength of a material volume element depends on
the "severeness" of the most critical flaw being present within that volume element.
The macroscopic strength distribution gel, x) of fibers of different length L but
uniform cross section can then be expressed as
45
x
gel, x) = Lnof(x)[1 - f f(y)dy]LnO-l (3.6)
where no is the number of flaws per unit length of fiber and f(x) the underlying
strength distribution. For f(x) being a Gaussian distribution with mean f.1 and
variance v, the most probable strength, a* , depends on Lno as shown by the solid
line (theory) in Figure 3.4. Also plotted are the experimental data of Levin and Savi-
ckij [10]. Even though it appears as if the experimental data would be fitted by
the theoretical curve in the range of 10 3 < Lno < 10 6 , some caution is required.
If indeed the inherent distribution of flaws is a Gaussian distribution, then the
decrease of a*(Lno) with L and the variance of a(Lno) are both related to v and
n(= Lno) in a known manner and subject to proof. A proof is also possible for other
underlying distributions. For the data of Levin and Savickij [10] neither a Gaussian
distribution of x, nor a Laplacian, nor a Weibull distribution describes a* (Lno) and
v(Lno) simultaneously [11]. As a result of this analysis one may conclude that
either the underlying distribution f(x) is truncated at the left-hand side (small values
of x), or the variance of a is much larger than to be expected from the number of
flaws. The truncation at small strength values could be caused by a multi-step crack
initiation process which would require a certain time. The effect of sample volume
on strength was also not in agreement with extreme value statistics for viscose fila-
ments, carbon fibers, and cotton yam [12]. Fedors and Landel [13] obtained a
doubly exponential distribution of strength values for eight SPR gum Vllicanizates.
Such a distribution derives, for example, from an underlying exponential distribu-
tion of flaw cross sections [14]. Here the volume effect was not studied.
When dealing with underlying distributions it should be recognized that dif-
ferent regions (core, surface) of a (seemingly) homogeneous specimen may have
different underlying distributions. Evidence of this behavior was for instance found
in polyamide fibers [15] and LOPE pipes [84]. Depending on processing conditions
volume-related and surface-related breakages with quite different stress-lifetime
characteristics could be distinguished. At this point the many attempts to determine
underlying distributions are not discussed any further but reference is made to the
publications on this subject [11-17]. If any conclusion may be attempted from
statistical investigations of uniaxial failure of thermoplastics it is that the underlying
distribution, f(x), of natural flaws could be a Laplacian distribution possibly skewed
at the left-hand side (small values of x). Moreover, it appears unlikely that a deter-
ministic view may be adopted associating a particular flaw with a particular strength
(or time to failure) but rather necessary to allow for the uncertainty of multistep
flaw growth.
The third statistical argument (fracture as the result of a large number of molecu-
lar processes) is very old. Its application to polymers definitely was stimulated from
many sides including fracture and fatigue of metals [4] or glasses [17] and the ther-
modynamics ofreactions [19]. The interpretation of fracture as a sequence of in-
dividual steps will be mainly used in the following. It necessitates a consideration
of the correlation of such steps and of the final failure criterion.
The discussion of the three different statistical arguments can be summarized
by saying that in the first case (fracture as a statistical event) a property (probability
46
III. Continuum and Fracture Mechanics Approach
of fracture) is attached to the material body as a whole. In the second case one
flaw, i.e., one micro-heterogeneity (out of many), within the body is considered
to dominate failure; in the third case many events interact, influence, and determine
each other. The very same lines of thought used here to explain the variability of
strength,for example, are found in the continuum mechanics, fracture mechanics,
and molecular approach to theories of strength, respectively.
A. Classical Failure Criteria
In classical continuum mechanics homogeneous isotropic materials are taken into

consideration. Failure criteria are established on the basis that there is indeed one
strategic material property such as uniaxial tensile strength, shear strength, elastic
or total extensibility, or energy storage capability that determines the failure of
the stressed sample. If, in establishing such criteria, all environmental parameters
(e.g., temperature T, strain rate, e, or surrounding medium) are taken to be con-
stant, then failure has to be expected whenever the components of multiaxial stress
- usually the three principal stresses 01, 02, and 03 are considered - combine in
such a way that the strategic quantity reaches a critical value C. For different Tor
eC may assume different values. The condition f(Ol, 02, 03) = C(T, e) represents
a two-dimensional "failure surface" in three-dimensional stress space. The stable
states of stress form a continuous body bounded by the failure surface against the
unstable stress points.
It has already been indicated that failure in polymeric materials may occur in
a number of phenomenologically different ways, e.g., as brittle fracture through
propagating cracks, as ductile fracture through plastic deformation following shear
yielding, or as quasi-brittle fracture after normal-stress yielding (crazing). It had
correctly been anticipated that these different phenomena respond differently to
the magnitude and the state of stress. This means that for different fracture phe-
nomena separate failure surfaces exist which may overlap and penetrate each other.
These facts are extensively investigated and discussed in the recent monograph by
Ward [20] and in the general literature (e.g., 21-34).
The classical criteria were conceived to describe the failure of isotropic, elastic-
plastic solids. The physical content of the six principal criteria is in brief the follow-
ing:
1. The maximum principle stress theory (Rankine's theory) states that the largest
principle stress component, 03, in the material determines failure regardless of the
value of normal or shearing stresses. The stability criterion is formulated as
(3.7)
In this case 0* is considered a basic material property which can be determined

from, for example, a tension test. In stress space the surface of failure according to
47
the maximum stress theory is a cube. A modified form of this theory allows for a
larger critical value a* if one of the stresses is compressive. Then the surface of failure
is a cube again, but with the center displaced from origin.
2. The maximum elastic strain theory (St. Venant's theory) states that incep-
tion of failure is due if the largest local strain, €3, within the material exceeds some-
where a critical value €*. The failure criterion, therefore, is derived as
(3.8)
where E is the isotropic Young's modulus and v Poisson's ratio. The term in square
brackets thus has to be smaller than an equivalent stress, a*.
3. The theory of total elastic energy of deformation was first proposed by Bel-
trami. Since under high hydrostatic pressure large amounts of elastic energy may be
stored without causing either fracture or permanent deformation, the elastic energy
as such seems to have no significance as a limiting condition.
4. Theory of constant elastic strain energy of distortion (Huber, v. Mises, Hen-
cky). The consideration of the poor significance of the elastically stored energy in
hydrostatic compression or tension led to the idea of subtracting the "hydrostatic:'
part of the energy from the total amount of elastically stored energy. Thus, it is
assumed that only the energy of distortion, W, determines the criticality of a state
of stress. For small deformations one obtains:
16+Ev W =[(01 - a2)2 + (a2 - a3)2 + (a3 - al)2] <2(a*)2. (3.9)
5. If a limiting octahedral shearing stress, T*, is postulated as a failure criterion

the same mathematical expression as in Eq. (3.9) is obtained with
(3.1 0)
In this case the limiting condition of small deformations does not pertain.
6. The Coulomb yield criterion states that the critical stress, T*, for shear de-
formation to occur in any plane increases linearly with the pressure, an, applied
normally to that plane:
(3.11)
Here the material constant TO is given by the cohesion of the material and /1 is
a "coefficient of friction". If friction is neglected the Tresca yield criterion is ob-
tained.
The above failure (or stability) criteria do not explicitly contain time as a
variable. One may apply them to rate sensitive materials, however, if one recognizes
that 0* and T* will depend on stress history. The classical approach of Eyring and
other rate theories will be discussed in Section IV of this chapter. The limits of the
applicability of the classical criteria and of their extension to anisotropic materials
are analyzed by Ward [20]. Also in recent years numerous papers (e.g., 21-28)
48
and review articles [29-34] have appeared which concern the geometry of failure
surfaces of polymeric materials. Multiaxial failure experiments on elastomers which
seem to corroborate the St. Venant criterion have been carried out [21] by Gent
and Lindley on endbonded penny-shaped samples of a natural rubber gum vulca-
nizate, by Ko on polyurethan rubber, by Oberth and Bruenner on castable elastomers
containing solid inclusions, and by Lim [22] on tubular samples of polyurethane
and of a butadiene/acrylic acid copolymer. Figure 3.5 depicts Lim's data and a
failure envelope for v = 0.37 resulting from the St. Venant criterion .
1.0
""':7' oa
.~
''''~
a
')/' x
if 0.5 aa I ·~Huber-Hencky-v. Mises
/-:~
.f
// I
I •
-1.5 f.-1.O -0.5 1.5
/f
f
f -0.5
-1.5
Fig. 3.5. Failure in multiaxial stress, 0 PMMA tubes (Broutman et aI., (23]), [', 6 PA tubes,
... buckling (Ely, (24]), x PUR tubes (Lim, [22]),. SBR membranes (Dickie et aI., (25])
- - - maximum strain failure criterion, - . - . - octahedral shear stress failure criterion.
Of the other theories, the Huber-v. Mises-Hencky criterion and the Coulomb
yield criterion have received the most experimental support. The failure surface in
the stress space described by Eq. (3.9) is a cylinder with its axis along the space
diagonal. Traces of this cylinder with the (01, 02)-plane are ellipses symmetric with
respect to the point of origin. One such ellipse is shown in Figure 3.5. The failure sur-
face corresponding to the Coulomb criterion is a cone. Its traces with the (a 1 , 02)-
plane are also ellipses but displaced with regard to the origin. Elliptic traces seem
to form the best available failure criterion in the biaxial yielding experiments with
tubular specimens ofPMMA [23, 33], PVC [26,27], PC [27] and ofPS, PE, PP, PA,
ABS, and CAB [33]. In the case of multi axial fracture without yielding (Trenn-
bruch) elliptic and parabolic traces are obtained which pertain to a cone or a para-
boloide in stress space [30, 33]. In any event there is no strict distinction between
the form of failure surfaces describing failure through flow and the form of the sur-
faces referring to failure through material separation (brittle fracture). This indicates
that in these isotropic polymers the initiation of failure was strongly influenced by
shear stresses irrespective of the ultimate material behavior. No unique 0* or f*
49
proved to be a critical quantity in the experiments of Dickie and Smith [25] on the
fracture of SBR in equal biaxial tension and of Dong [28] on a highly filled fluor
elastomer (Viton ®).
More recently there has been a strong interest in the deformation and fracture
behavior of plastics under large hydrostatic pressures [31-32]. One should expect
- and one observes - that the rigidity of a polymer increases with pressure. Sauer
et al. [32] report that a pressure of 3.5 kbar raises the initial Young's moduli of
amorphous thermoplastics (PC, PI, PSU, PVC, CA) by a factor of 1.2 to 1.9, that
of crystalline polymers by 1.4 (POM) to 7.5 (PUR). Despite the increased rigidity,
ductile fracture occurs. The effects are not yet understood in all generality. Follow-
ing the two major review articles on this subject by Radcliffe [31] and Sauer and
Pae [32] the Coulomb criterion corresponds best to most pressure-yield experiments.
In the case of PTFE and PC with their nonlinear increase of yield strength with
hydrostatic pressure (an = p) even the consideration of a second-order term of p
appeared to be necessary [32]. Whereas the application of pressure restricts the cold
drawing of many ductile polymers, it enables yielding of normally brittle polymers,
e.g., of PS at room temperature at a pressure of 2 to 3 kbar, and of PSU and polyimides
at 3 to 7 kbar [32].
In this chapter no molecular interpretations of fracture phenomena are given.
The above discussion has already shown, however, that a consideration of the three-
dimensional state of stress is not sufficient to elucidate the possible role of chain
scission or disentanglement in failure of polymers. This is particularly true if the
phenomenon of crazing ("normal stress yielding") is included. Before investigating
the molecular aspects, however, in the later chapters, the presentation of general,
non-molecular fracture theories should be continued.
B. Fracture Mechanics
Real solids are inhomogeneous. Even if they are one-phase materials they will con-
tain defects, voids, inclusions, flaws, cracks, and/or other inhomogeneities which
are able to distort an otherwise homogeneous stress field. A continuum mechanical
approach that analyzes these (singular) stress-strain fields in the vicinity of flaws
or cracks and derives its criteria of crack stability from an energy balance considera-
tion is the fracture mechanics approach. Griffith [35] was the first to find wide
acclaim for his considerations of the energetic changes occurring as a consequence
of the widening of a crack (of length 2a). He equated the resulting decrease dU of
elastic energy stored within a stressed specimen with the surface energy 'YedA
necessary to enlarge the crack by an infinitesimal area dA:
dU dA
- - ='Y - (3.12)
da e da·
For a plate containing an elliptical hole and stressed at right angles to the major
axis a, he obtained the stability criterion
0* = (2'YeE/na)1/2. (3.13)
50
The size of the largest material defect together with the material constants
modulus of elasticity (E) and surface work parameter (1c), thus, supposedly deter-
mine the strength of the material, Le., that macroscopic value a* of uniaxial stress
at which irreversible crack propagation begins. Equation (3.13) gives a mathematical
form to the earlier used concept that flaw size (or severity) determines the inherent
strength of a sample. It also explains that the actually observed macroscopic strength
is much smaller than the theoretical strength of flawless specimens.
The application of fracture mechanics to viscoelastic media is caught between
the limitations that are brough about by the deviation from infinitesimal strain con-
dition, by molecular anisotropy, local concentration of strains, and the time depen-
dency of stress and strain functions. It has been notably successful in investigations
of crack propagation. The analytical extension of Griffith's work to linearly vis-
coelastic materials has been reviewed by Williams [36] and more recently by Knauss
[37]. In Chapter'9 a more detailed account on the application of fracture mechanics
to crack propagation is given. There the morphological aspects and the effect of
time-dependent plastic deformation, craze initiation and growth, and chain scission
on the energy of cohesive fracture of high polymers will be considered.
At this point attention will be drawn to an application of the Griffith approach
to the effect of matrix orientation on strength. Such an interpretation has been
forwarded by Mikitishin et al. [79] and Stemstein et al. [80]. These authors suggest
that during the orientation process a rotation, elongation, and widening of possible
defects takes place. The change of shape and orientation changes the effect of these
flaws as stress concentrators. They become less dangerous in a direction parallel to
the draw direction (z) and more critical in the perpendicular direction (x). If only
the increase in length of (elliptical) voids which are oriented parallel to the draw
direction is considered, the lateral strength ax is derived from Eq. (3.13) as:
(3.14)
where Ais the draw ratio. Sternstein et al. [80] have made comprehensive calcula-
tions in two dimensions of the flaw-rotation and widening effect on axial and lateral
strength for a statistical distribution of flaw orientations and various flaw geometries.
They compare their predictions of the lateral strength of polystyrene with the ex-
perimental data of Retting [81]. For moderate molecular weights and draw ratios
up to 1.5 they find a correspondence to within 20%. For higher molecular weights
and larger draw ratios the discrepancies are above 70%.
If a crack travels at the interface of an adhesive joint one observes an adhesive
debonding rather than cohesive failure. Except for minor differences in boundary
condition the analgy to the fracture mechanical treatment of cohesive failure is
complete [38-40] and the normal debonding stress a: is obtained as:
(3.13a)
The interpretation of 1a and E is rather involved. These quantities not only

depend on the static surface free energies and elastic moduli of the interacting
(polymeric) materials but also on the work of deformation which the adhesive layer
51
undergoes in debonding and on the surface activity of a possible third phase having
access to the debonded area [40].
C. Continuous Viscoelastic Models
Under the heading of this section belongs the very fruitful model of a common
"failure envelope" proposed by Smith [41] It is based on a consideration of the
viscoelastic behavior of continuous high polymer bodies, i.e., on the concept that
a reduction according to the time-temperature superposition principle of the en·
vironmental parameters stress, strain-rate, and temperature should lead to corre-
sponding molecular states. If a fracture criterion refers indeed to unique limits of
molecular load-carrying capability, for example, then the plotting of reduced stress
at break versus strain at break for different experimental conditions should lead
to one master curve, the failure envelope (Fig. 3.6). For a large number of natural
and synthetic rubbers and vulcanizates under similar modes of mechanical excita·
tion, this concept was found to hold. It thus generally allows the extrapolation of
the ultimate behavior into unknown regimes of time, strain·rate, or temperature. This
model, however, does not allow prediction of the shape of the fracture envelope or
1.5
Symbol Temp °c
~ .
.. " ,, '\
N X no
100
""~
E
.......
0
1.0 ~ 70
Z "Q '0
~ 0 25
a 10
EI~ ..
II -5
-20
-35
tl
.Q. 0.5 "
"tI' - ~5
0\
0
-0.5
-0.2 o 0.2 0 .4 0 .6 O.B

log (A.b- 1 1
Fig. 3.6. Fracture envelope: reduced breaking stress vs. breaking elongation for an unfilled butyl
rubber vulcanizate (after Smith, (411).
52
IV. Rate Process Theories of Fracture
the conversion of data obtained under different modes of excitation (uniaxial,

equal biaxial, or other multiaxialloading).
An argument to resolve the discrepancy between the failure envelopes obtained
for different modes of straining is indicated by the work of Blatz, Sharda, and
Tschoegl [42]. These authors have proposed a generalized strain energy function
as constitutive equation of multi axial deformation. They incorporated more of the
nonlinear behavior in the constitutive relation between the strain energy density and
the strain. They were then able to describe simultaneously by four material con-
stants the stress-strain curves of natural rubber and of styrene-butadiene rubber in
simple tension, simple compression or equibiaxial tension, pure shear, and simple
shear.
The continuous viscoelastic models discussed in this section make use of the
theory of rubber elasticity and of the time-temperature superposition principle.
They implicitly very well recognize the molecular origin of the viscoelastic behavior
of a material - but explicitly they do not refer to noncontinuous quantities such
as the discrete size, structure, and arrangement of molecules, the anisotropy of
molecular properties, or the distribution of molecular stresses or stored strain energy.
If individual molecular events or discontinuities of stress or strain are either not
discernible or not of particular importance, a representation of a solid as a contin-
uum is quite adequate.
A. Overview
As opposed to the continuum mechanical theories, "molecular" rate process theo-

ries of fracture recognize the presence of discrete particles or elements forming
the material body. Rate process theories intend to relate in a straightforward man-
ner breakage, displacement, and reformation of these elements to deformation,
defect development, and fracture of the structured material.
All theories of a first group to be described subsequently have in common the
assumption that macroscopic failure is a rate process, that the basic fracture events
are controlled by thermally activated breakages of secondary and/or primary bonds,
and that the accumulation of these events leads to crack formation and/or to break-
down of the loaded sample. Here the basic events are defined in a general manner
and not experimentally related to particular morphological changes. A second group
of theories is based on the physical evidence of molecular damage obtained by
spectroscopy and X-ray scattering methods it will be discussed in Chapters 7 and
8. A third group of theories concerns the role of cracks and craze initiation and is
treated in Chapter 9.
One of the early applications of the theory given by Glasstone, Laidler, and
Eyring [43] was the prediction of the flow and diffusion behavior of liquids. From
the original concept of flow, the thermally activated jump of molecules across an
energy barrier, various fracture theories of solids have emerged. Tobolsky and
53
Eyring [44] considered the general decrease of secondary bonds, Zhurkov [45-47]
and Bueche [48-50] that of primary bonds to be the strength controlling factor.
In their recent monographs on deformation and fracture of polymers Bartenev and
Zuyev [51], Andrews [52], and Regel, Slutsker, and Tomashevskii [53] have re-
viewed a considerable amount of experimental data under this aspect; reference
to their works should be made if information is sought on the effect of time and
temperature on the fracture behavior of polymers of different composition and
structure and under various environments.
Following the earlier works of Eyring, Zhurkov, and Bueche, these and other
authors have continued to develop different aspects of the kinetic theory of frac-
ture. Particularly in the USSR fracture data have been interpreted in terms of the
failure of regular, nonmorphological model lattices (e.g., 54-58). Gubanov et al.
[54], Bartenev [55], and Perepelkin [56] focussed on the potential energy of inter-
action between adjacent members of a polymer chain, Mikitishin et al. [57] on the
defects introduced into the lattice by chain ends, and Dobrodumov [58] on the
increase of the load and - consequently - of the rate of scission of bonds adjacent
to a broken bond.
Hsiao and Kausch [59 -61] studied the effect of orientation -dependent local
strain on the total rate of chain scission. Holzmillier [62], Bartenevet al. [63], and
Salganik [64] analyzed the amount of thermal energy and the direction of relative
motion imparted by statistically fluctuating phonons to adjacent segments. Dif-
ferent statistical aspects of the accumulation of molecular defects were treated by
Orlovet al. [65], Goikhman [66], and Gotlib [67] who simulated the isolated pro-
duction, growth, interaction, and coalescence of defects. An energy probability
theory was forwarded by Valanis [68] who combined the role of strain energy den-
sity, the stochastic nature of fracture, and the fracture hypothesis of Zhurkov.
Whereas all of the above theories [54-68] refer in one way or another to the
thermally activated breakage of one kind of molecular bonds as primary fracture
event, the Bueche-Halpin theory for the tensile strength of gum elastomers [69]
employs the idea that the viscoelastic straining of rubber filaments together with
the degree of their ultimate elongation determine the kinetics of crack propagation
in an elastomeric solid.
In the following the characteristic aspects of these generalized, nonmorphological
rate process fracture theories are to be discussed, i.e. the nature of the basic fracture
events, their distribution in space and time, the law of the accumulation of these
events, and resulting fracture criteria.
B. Eyring' Theory of Flow
Flow of a liquid, melt, or solid is the result of a stress-biased thermodynamically

irreversible displacement of molecules past each other. A molecule in thermal
equilibrium with its surroundings is in thermal motion, which in the case of a liquid
or solid is predominantly a vibration around a temporary equilibrium position. The
amplitude of vibration is constantly changing. Eyring [43] postulated that a displace-
ment ( or jump) of a molecule from an initial to a neighboring equilibrium position
54
may occur if the thermal energy of the molecule is sufficient to reach the "activated"
stage, i.e., the top of the energy barrier separating the initial and the final equilibrium
positions. The rate of decay of activated states towards the final position was ob-
tained by him as
kT
ko = t<. h exp (-uo/kT) exp As/k. (3.15)
Here t<. is the transmission constant which indicates how many of the activated
complexes actually disintegrate, k is Boltzmann's, h Planck's, and R the gas con-
stant, T is absolute temperature, Uo the height of the potential barrier, and As the
difference in entropy between ground state and activated state (the quantities u
and s in small letters refer to one particle, the capital quantities to a mole of them).
In the absence of external forces initial and final equilibrium states are assumed to
have the same potential energy. Then the rates of flow of particles across the divid-
ing potential barrier are the same in the forward and backward directions. Under
the influence of external forces, however, local stress fields are generated. A particle
thermally activated to move in the direction of, say, stress qr by a distance 'A/2 then
gains an amount of energy w:
(3.16)
where q is the average cross section occupied by the particle normal to the direction
of motion. A particle moving in the -qr direction will loose the same amount of
energy. The first particle, therefore, needs less thermal energy to reach the activated
state. than the second particle. Consequently the rates of flow of particles across an
energy barrier in opposite directions are biased by the local stress:
kl =ko exp (w/kT) (3.17)
k2 = ko exp (-w/kT). (3.18)
The net flow rate, therefore, is:
K = kl - k2 = 2 ko sinh (w/kT). (3.19)
It should be noted very carefully that w is equal to the product of local stress
times an activation volume; this is the volume drawn up by particle cross section
and step width in the process of activation. Activation volume and activation energy
necessarily must refer to one and the same activation process.
c. Tobolsky-Eyring
The concept of slipping of secondary bonds was applied by Tobolsky and Eyring
in 1943 to the breaking of polymeric threads under - uniaxial - load [44]. The
55
rate of decrease of the number N of such bonds per unit area under constant uniaxial
stress 'lro is obtained from Eqs. (3.1) and (3.l9):
-! dN = 2 ko sinh ('Ir 0 A/2 NkT). (3.20)

N dt
For large values of stress the flow of bonds takes place almost exclusively as
breakage and not as reformation. In that case one may use Eq. (3.17) instead of
(3.19). If one substitutes there 'lroA/2 NkT by y one obtains a solution in terms of
the exponential integral -Ei( - y):
00
-Ei(-y) = f dy'exp(-y')/y' =kotb. (3.21)

Yo
The lower limit of integration isyo = 'lroA/2 NokT = w/kT. Equation (3.21)
offers an implicit relation between the expectation value of the time to fracture of
the thread, t b , the rate of flow of unstressed secondary bonds, ko, and the stress
'lro acting on the No initial bonds. With w ~ kT the exponential integral may be
approximated by:
-Ei(-y) = [(exp - y)/y](1- l/Y)· (3.22)
Within the range of validity of this approximation the logarithm of lifetime,

log t b , is almost linearly related to w, i.e., to the applied stress. Exactly this behavior
is exhibited by a large number of stressed metals, ceramics, or polymers. Together
with Eyring's original interpretation this has laid the foundations of the kinetic theory
of fracture, to which - as indicated above - subsequently a large number of research-
ers have contributed.
D. Zhurkov, Bueche
Independently from each other Zhurkov et al. [45-47] in the USSR and Bueche
[48-50] in the USA expressed the idea that in fracture of polymers - and also of
metals for that matter - the breakage of primary (chemical) bonds plays an im-
portant role. They found that the time to fracture under uniaxial stress ao of Zn,
AI, and for example, PMMA and PS below their glass transition temperatures could
be expressed by an exponential relationship involving three kinetic parameters:
(3.23)
The three parameters were - based on the available data - interpreted as energy
for activation of bond scission (U o), as the inverse of molecular oscillation frequen-
cy (to), and as a structure sensitive parameter 'Y. In Figure 3.7 experimental data and
the theoretical curves according to Eq. (3.23) are reproduced.
Equation (3.23) can be considered as a general expression of the kinetic nature
of material disintegration. A closer discussion of the meaning of the molecular os-
cillation frequency, of the structure-sensitive parameter 'Y, and also of the activation
56
o
-5
-10
o
Fig. 3.7. Times to fracture under constant uniaxial load: cellulose nitrate (CN) at T = 70 (1),
30 (2), -10 (3), and -50°C (4); polyamide 6 (PA 6) at T = 80 (1), 30 (2),18 (3), - 60 (4),
and - 110°C (5) (after Savitskii et al. (86).
energy for bond scission in view of the molecular structure and of the sequence of
molecular processes will be given in Chapters 7- 9. At this point one would like to
check further arguments that the existence (and volume concentration) of the strong
primary bonds determines the rigidity and strength of (polymeric) solids.
Holliday et al. (70] have looked under this aspect on the degree of connectivity
CN (average number of main network bonds per chain atom, e.g., 2 in hydrocarbon
chains, 3 in graphite , and 4 in diamond) and on the relative volume density of main
bonds NCR (which here is density times number of network atoms in the repeat
unit divided by molecular weight of a repeat unit). The term NCR varies between
less than 0.01 for bulky polymeric chains and 0.58 for diamond. The bulk moduli,
Young's moduli, hardness values, and cubical coefficients of thermal expansion of
the more than 50 organic and inorganic polymers investigated at room temperature
exhibit a strong correlation with NCR. In Figure 3.8 the best fit curves drawn by Holli-
day for Young's modulus and Brinell hardness values BHN are reproduced (the ac-
tual NCR values - except for those of the graphitelike structures - fall into a narrow
band along the curves having a width of ± 0.05). The nearly identical shapes of the
upper portions of both curves suggest that hardness (compressive yield strength)
and elastic modulus depend in the same way on the concentration of primary bonds.
In fact for the points on the curve with NCR> 0.08 (mostly organic thermosetting
resins or inorganic polymers) the ratio of BNH value and E has a constant value of
0.0625.
For values of NCR < 0.08 (thermoplastic materials) the correlation between
mechanical properties and bond density is much less pronounced although the steep
slopes of the shown curves tend to make the scatter of the actual data points less
apparent. The main reason for the lack of good correlation certainly is the fact that
in high polymer solids the stiffness and strength can be utilized only to that extent
permitted by intermolecular attraction and segment orientation (cf. Table 1.1 for
the differences in elastic moduli of chains, highly oriented fibers, and isotropic
solids; in section E below and in Chapter 5 the role of orientation and intermolecular
attraction on chain strain is discussed in detail).
In order to investigate the effect of the density of main chain bonds on the
mechanical properties it is necessary to select experimental conditions such that
57
E BHN
lOs
10 6
MN/m2 SiC )(
~nd MN/m2
10'
lOs
-- - E
BHN
.'"
"
"0
>
'"
10 3 ~
c
10' '0
'" ::;
'"c
L;
PS
"0
>- x
-;
c
66 PA ~
10 2
10 3
I X PE
I
x PTFE
10'
10 2
0.2 0.4 0.6 NCR
relatl've number of pnmary network bonds
Fig. 3.8. Young's modulus E and BrinelJ hardness values BHN of various solids vs. volume con-
centration of primary bonds (after Holliday [70]).
comparable mechanical excitation of said bonds does indeed take place. According
to Vincent [71] commensurable reference temperatures to measure a critical tensile
strength are the brittle point temperatures, where brittle failure changes to a ductile
failure. As indicated in Figure 3.9 he has related the number of backbone bonds per
unit area (i.e., sample density times length of repeat unit over weight of repeat unit)
to the critical tensile strength ()f (isotropic) polymers. A linear relationship exists
between critical sample strength and bond packing density which does not seem to
depend on the chemical nature of the bond. From the slope of the straight line the
critical strength (in MN/m 2 ) of the isotropic polymers is obtained as 36.8 times
the number of backbone bonds per nm 2 . In highly oriented textile fibers this figure
is much larger and amounts to between 100 and 250; in aromatic polyamide fibers
(Kevlar 49) it reaches about 650. This should be compared with the theoretical
value for fully oriented chains of infinite length which is about 3000, since the axial
load to break a molecular chain of anyone of the species listed in Figure 3.9 is of
the order of 3 nN.
In other words, in isotropic thermoplastics at the point of brittle failure only
a minute fraction - less than 1% - of all main chain bonds is fully strained. Under
these conditions, onset of unstable crack propagation is determined by the mag-
nitude of intermolecular attraction. Reinterpreting Vincent's data one must say that
it is not the number of backbone bonds per unit area and their accumulated strength
58
o POM
200
MN/m2
PA
160 0 PE
PET
PC PES
PVC
Q
.c.
c
..
0.
t; 120
.!!
PTFE
1/1
C
.2! PP
.,
C
.u
'':
80 PB
u
PMMA
PPe
P4MP
60
Fig. 3.9. Critical tensile strength

0 (strength at ductile-brittle transi-
0 2 I. 6 tion) vs. chain cross-section (after
number of backbone bonds per nm 2 Vincent [71 D.
which determine macroscopic strength but the level of intermolecular forces; the
latter benefit from a closer packing.
The effect of molecukIr weight on fiber strength has been widely investigated
(cf. 20, 51, 52, 54, 72-74). The general observations have been briefly summarized
and illustrated by Figure 1.13 for PA 6 fibers. In low molecular weight material chain
slip occurs readily so that the sample strength depends solely on the strength of
intermolecular attraction. Noticeable macroscopic strength is only obtained if the
molecular weight is sufficient to permit physical cross-links through entanglements
or chain folds between several chains.
In the following section it will be discussed in more detail to what extent the
orientation distribution of chain segments, the ensuing distribution of axial chain
stresses, and the gradual accumulation of defects through successive breakage of
chains can be held responsible for the discrepancy between actual and theoretical
strength.
E. Hsiao-Kausch
In the fracture theory of Hsiao-Kausch [60, 61] the state of orientation of a polymeric
solid is explicitly recognized. The theory combines the kinetic concept of Zhurkov or
Bueche and the theory of deformation of anisotropic solids developed by Hsiao [59]
59
for a model made up of rodlike elastic elements. The fracture theory is based on the
simplifying assumption that the mechanical properties of the anisotropic solid are
predominantly determined by the state of orientation and the properties of the (one-
dimensional) elastic elements. The intermolecular interaction is neglected. In Chapter 2
the mathematical model (Fig. 2.12) has been described and the constitutive stress-
strain relations (Eq. (2.5)) have been derived. The kinetic aspects are introduced into this
model through the assumption that the rodlike elastic elements can undergo breakage,
that the probability of failure of an element is determined by the axial stress acting
upon that element, and that an immediate redistribution of the load carried up to
then by the breaking element occurs. Again the models is characterized by the fact
that it distinguishes between the macroscopic state of stress (fJ) and the microscopic
(axial) stress 'I! of an element. The axial stress of a one-dimensional elastic element
deformed in a homogeneous strain field Emn is determined by its axial elongation:
(3.24)
The summation convention for repeated indices is used. E is the modulus of

elasticity of the elastic elements, and sm, Sn are components of the unit vector in
the direction of orientation identified by {} and cp (S1 = sin {} cos cp, S2 = sin {} sin cp,
s3 = cos {}). For uniaxial stress, 033 = 00, and transverse symmetry about the
33-axis Eq. (3.24) becomes:
(3.25)
where II is Poisson ratio Ell /E33.

In a partially or nonoriented system, the microscopic axial stresses are larger
than 00 by a factor of up to E/E"A. where E"A. is the longitudinal modulus of elasticity
of the subvolume containing the elastic element. If all subvolumes of a sample behave
identically, then E"A. is the longitudinal modulus of the (partially) oriented sample.
Naturally the largest axial stresses are experienced by those elements oriented in
the direction of uniaxial stress. In a completely oriented ~ystem E and E"A. are equal
by definition, and macroscopic and microscopic stress (in the direction of orienta-
tion) are equal also.
As stated above it is then first assumed in accordance with Zhurkov's kinetic
considerations that elements break and secondly that the orientation-dependent
rate of breakage of elements is given by Eq. (3.17) as
d f ({) t)
- ' = k ({}) = ko exp (f3'1!/RT) (3.26)
fdt
where f3 is the activation volume for breakage of the elastic elements in question
and f= N({}, t)/N({}, 0) the fraction of unbroken elements.
Thirdly it is assumed that after breakage of an element mechanical equilibrium
is established within the affected subvolume and that strain E33 increases. This means
that the elastic elements of all orientations take over an appropriate share of the
load carried previously by the broken element to balance the load acting on the
60
boundaries of the sub volume. It has been shown in Chapter 2 that the stress tensor
acting on a small volume element of the elastic network can be calculated once the
orientation distribution function, p, is known
Ojj = J p(&) f(~, t) Sj Sj 'l'(~, t) dn (3.27)
where dn is the infinitesimal solid angle.

In uniaxial stressing, mechanical equilibrium within the elastic network is
established if
n/2
011 =022 =0 =J Ep(~) f(~, t) [€ll sin 2 ~ + €33 cos 2 ~]sin3 ~d ~ (3.28)
o
and
n/2
033 = 21T J Ep(~) f(~, t) [€ll sin 2 ~ + €33 cos 2 ~] cos 2 ~ sin ~ d~. (3.29)
o
The system of Eqs. (3.26), (3.28), and (3.29) has been evaluated for a complete-
ly oriented and an unoriented network under constant uniaxial stress 00. The case
of a completely oriented network is that treated by Tobolsky and Eyring. All ele-
ments are thought to be subjected to the same stress 'l' which increases inversely
proportional to the decreasing number of unbroken elements. Breakdown of the
subvolume occurs with breakage of the last element within the subvolume, i.e.,
after f has reached zero. The time to breakdown of the subvolume, tb, is taken from
Eqs. (3.20), (3.21), and (3.26) as
koto = - Ei (-iN'0 /RT) (3.30)
where'l'o is the initial stress carried by elements of orientation ~ = O.
In the case of random network orientation the system of Eqs. (3.26), (3.28),
and (3.29) cannot be solved analytically. An approximation method has been pro-
posed, however [60]. Without repeating details of the method it may be said that
the changes of f for a time interval Ll t as calculated from Eq. (3.26) are being fed
back into Eqs. (3.28) and (3.29) giving the corresponding changes of €ll (t) and
€33(t), which in turn determine the changes of 'l'(t). The iterative integration of
Eqs. (3.26), (3.28), and (3.29) can be carried out numerically.
Two results of these calculations seem to be of interest with respect to the
effect of network orientation on defect accumulation and strength: that the range
of orientation angles within which elements break preferentially is narrow and that
the increase in strength resulting from improved uniaxial orientation is limited. The
first effect is illustrated in Figure 3.10 for a randomly oriented network where the
initial distribution p = 1/2 1T, of elements is independent of~. Elements oriented
in the direction of uniaxial stress (cos ~ = 1) experience the largest initial stress
('l' 0), and they break preferentially. The angular distribution of f (cos ~, TIME) is
shown for three different stages of defect development, namely after f (1, TIME)
has dropped to 0.5 and zero, respectively, and for the point of impending network
failure - where the numerical calculations had been terminated because breakage
61
_I_ II
TIME= 038 060 0633
(j)
I-
z w
w ::;:
::;: i=
w
-'
w
..,-
(j)
z 0 ~<l'.'RT = 25
w ~
}lu
~
0
(l:: 05 1.0 10 10
£D
z
:::J TIME= 044
LL
0
(l:: 05
w
£D
::;: ~<jJ./RT = 3
:::J COS.;}
z
0 05 10 10 1.0
Distribution of Elements
Fig. 3.10. Decay within a population of randomly oriented elements; at high stresses
({3'l10 > 10 RD essentially those elements oriented in the direction of uniaxial stress break
(upper graph). TIME = linear time scale according to Eq. (3.31) (Kausch, Hsiao [60]).
of the .remaining elements occurred at too high a rate. For ease of comparison the
time scale is given in terms of TIME where
TIME = t/t z =ko t exp ({3'1t 0 /RT) (3.31)
and t z the time to failure according to Zhurkov's Eq. (3.23). It should be noted
that more than one-half of the loading period has to pass before one-half of the
most heavily stressed elements is broken. More than 90% of the loading period has
to elapse before all elements within a small angular section are broken. At this
second stage the majority of all elements is still unbroken. The step from the second
to the third stage of defect accumulation is rather short « 6% ofloading time) and
from there to network failure practically negligible. It should also be noted that
TIME (failure) of a randomly oriented network composed of successively breaking
interacting elements is only somewhat smaller than the time given by the empirical
Eq. (3.23) where TIME (failure) for a stress 'Ito is unity; on the other hand it is
larger than RT/{3'1t, the value predicted by the exponential integral (Eq. (3.22)) for
a completely oriented network loaded to such an extent that again the elements
experience the same initial stress 'Ito.
The numerical calculations according to the above model resulted in an estimate
of the time to failure of arbitrarily oriented networks under constant load:
ko tb = -[Ei (-{3'1to/RT)]C (3.32)
where C is a slowly varying correction term the limits of which are expressed by:
1 ~ C"::;; 1 + 0.63 {3 'lto/RT. (3.33)
The lower limit refers to completely oriented, the upper to randomly oriented
networks. With C being determined it has become possible to compare Eqs. (3.23)
62
and (3.32) and to express the structure sensitive kinetic parameter -y in terms of
the element properties. In the range of validity of Eq. (3.22) one obtains for a fully
oriented network:
'110 ( 1 -RT
-y=IJ- - l nRT)
- (3.34)
00 lJ'I1o lJ'I1o
and for a randomly oriented one
-y = IJ -'110 (1R
- -T In 0.63 ) . (3.35)
00 1J'I10
The term in parentheses is an orientation-dependent and minor kinetic contri-

bution derived from the assumption of successive failure of many elements (in the
technically relevant range of 1J'I10 values from 40 to 200 kJ /mole the term in paren-
theses assumes values between 1.17 and 1.0). If the nature of the breaking elements
is not changed during sample treatment or fracture development, then IJ can be con-
sidered to be constant. The major effect on -y is derived from the local stress con-
centration 'I1o/ao which here is equal to the stiffness ratio E/E?-. This theory essen-
tially predicts, therefore, that the increase in strength is equal to the increase in
stiffness; it is based on the presumption that the elements break indeed at a critical
local strain (kinetic version of St. Venant's criterion of maximum strain). A different
interpretation of the strength of oriented samples which is based on the rotation of
flaws has been presented in an earlier section of this chapter.
In Figure 3.11 the experimental data obtained by Regel and Leksovskii [75]
for the lifetimes of partially oriented PAN fibers are compared with theoretical
curves derived from Eq. (3.32). It should be emphasized that the gain in strength
uniaxial stress
Fig. 3.11. Time to fracture of differently oriented PAN fibers. - experimental curves
(Regel et al. (75)), ---- theoretical curves (Eq. 3.32), parameter: draw-ratio "A.
through improved orientation of the PAN fibers (or of their model representation)
amounts to a factor of '110/00 = 5. Similar values of between 2 and 5 for the increase
in strength with orientation have been reported for PE, PP, PS, PVC, PMMA, or PA
[51, 54]. It should be pointed out that the rather limited increase in stiffness in
63
these experiments indicates that the orienting "elements" are not simply highly ex-
tended chain segments but larger molecular domains of limited anisotropy. This
does not preclude the possibility that breakage of an element is essentially identical
to breakage of the most highly stressed chain molecule(s). This will be the case if
the breaking chain(s) carry practically the total load of an element or initiate its
breakage.
A comparison of different reaction rate models was made by Henderson et al.
[76] who found that the bond rupture theories (Tobolsky-Eyring, Zhurkov-Bueche)
gave better agreement with fracture data on filled poly butadiene and filled and un-
filled PVC than the bond slippage hypothesis. Primary uncertainties exist, however,
with respect to a repair reaction and to the temperature dependence of {3 and 'Y- The
question of the true meaning of the activation volumes {3 and 'Y, of the stress ratio
wo/ao, and of the role of chain stretching and chain scission in macroscopic frac-
ture will be resumed in Chapters 7 to 9.
F. Gotlib, Dobrodumov et al.
In none of the previously treated molecular fracture theories (A - E) allowance has

been made for possible stress concentration in the neighborhood of a breaking
element. The first ruptures in a large ensemble of initially equally stressed molecules
certainly will occur at random within the stressed volume. The breakage and retrac-
tion of a particular element must lead, however, preferentially to an increase in
axial stresses of those elements to which it is coupled (through secondary forces).
The probability density for rupture of those elements will, therefore, be slightly
higher than that for others. But the higher probability density applies only to the
small number of elements in the vicinity of the existing fracture points. So the (in-
tegrated) probability for rupture of one of these elements is initially smaller than
the probability for rupture of one of the large number of chains hitherto not af-
fected by a ruptured element. The rupture events, therefore, will continue for some
time to occur at random [67]. With their growing number, chances are increasing
that rupture events occur in an immediate vicinity of each other thereby forming
the nucleus of a crack and enhancing the fracture probability of neighboring
elements.
Whenever the random accumulation of defects in one site has grown to such
a size that subsequent fractures will preferentially occur at that site, one has reached
the end of the phase of fracture initiation and the beginning of (thermal) crack
growth
In Figure 3.12 a computer simulation by Dobrodumov et al. of the fracture
development within a regular model lattice is shown [58]. The authors have taken
the ratio of the elastic constants in and perpendicular to the direction of orientation
as lO: 1 and an element fracture probability as given by Eq. (3.23). The distribution
of local stresses was determined by a numerical iteration procedure. It is readily ob-
served that for small loads ({3w o /RT = 5) a large number of isolated defects appear
before (from the accidental accumulation of a few defect points) a "crack" starts
growing. At higher loads ({3w o /RT = 20) a smaller number of defect points initiate
64
mode l '
n ·
• 1
~'l
I·' ' .. ,.'P" <.'J 1--',
l t
~:1 _"',
.~ . ._,'J I· IUtI" "'I" I
. ... '''\' '.1
I
{Jl{/o
RT
I . .
. , . .', 5
1 . . . ._ _ . ~.
I Ii " . ...' !I. ~.~.'
"
I·
i.~ i:::.....2....:J
Fig. 3.12 a and b. Simulation of brittle fracture of a three-dimensional network model (after [58 D·
Upper part : model representation, elastic constants Kl :K2 = 10. Bottom part: successive frac-
ture events in x-y plane as a function of tlr (r = time to complete failure); note the increase in
cooperativeness with {3wo/RT similar to that in Figure 3.10.
a crack and very little damage is done to the remainder of the plane during fracture
initiation. These results are qualitatively similar to those of Hsiao and Kausch shown
in Figure 3.10. From these studies the authors [57, 67] conclude that in polymers
only weak stress concentration on a molecular scale can exist (axial stresses adjacent
to a site comprising j defects should be larger by a factor of VI'than the stress further
away). It should be noted that this model - as the previous one - strictly operates
with elastic elements and with their rupture as the only way to relieve local stress.
The application of these models, therefore, seems to be restricted to the brittle frac-
ture of organic glasses or to other systems not involving large anelastic or plastic
deformation in fracture initiation.
G. Bueche - Halpin
The limited viscoelastic extensibility of rubber strands determines according to the

theory of Bueche and Halpin [69] the fracture of elastomers. The authors assume
65
that a large number of filaments at the tip of a growing crack have to be strained
successively to their critical elongation Ac. The sample will fracture at a gross strain
Ab after q filaments have been broken in a time tb = qt'. The values Ab and Ac are
connected through the creep function of the material and the stress concentration
factor. The theory allows one to calculate the elongation at break, Ab, from the
knowledge of the creep function. It does not take into consideration any change
of stress concentration with growing crack length or the decrease of t', the time
necessary to fracture one filament, with continuing creep of the sample. It is as-
sumed that all filaments will have to be stretched from practically zero elongation
to Ac. That will, in the first place, affect any numerical valUes of q which may be
calculated from fitting experimental failure envelopes. A group of q filaments
subject to the statistical condition that fracture of one filament may start once
the fracture of the preceding one is completed has an average lifetime, < tb >, of
qt' and a Poisson distribution oftb:
(3.36)
The same concept of filament deformation was independently employed by

Bartenev et al. [77] to the opening and growth of crazes (silver cracks) in PMMA.
The molecular theories treated in this chapter have considered the thermally
activated breakage of elements or filaments as a source of crack initiation and
macroscopic failure. It is the purpose of the following chapters to investigate the
mechanical strength of primary bonds and of molecular chains and to study the
occurrence of chain breakages. With that information available it will be possible
to resume the discussion on the nature of the "elements" used in molecular theories
of fracture.
1. R. M. Bethea, B. S. Duran, T. L. Boullion: Statistical Methods for Engineers and Scientists,

New York: Marcel Dekker 1975.
2. A. Hald: Statistical Theory with Engineering Applications, New York: Wiley 1952.
3. H. H. Kausch: Materialpriif.6, 246 (1964).
4. T. Yokobori: Kolloid-Z. 166, 20 (1959).
5. S. Kawabata, P. J. Blatz: Rubber Chern. Techno!. 39,923 (1966).
6. S. Kawabata, S. Tatsuta, H. Kawai: Proc. Fifth Internat. Congr. Rheology, Kyoto, Japan 3,
111 (1968).
7. B. D. Coleman: J. App!. Phy. 29, 968 (1958).
8. B. D. Coleman, D. W. Marquardt: J. App!. Phys. 28, 1065 (1975).
9. B. Epstein: J. Appl. Phys. 19, 140 (1948).
10. B. Ya. Levin, A. V. Savickij: Chim. Volokna 5,48 (1968).
11. H. H. Kausch: Proc. Internat. Conf. Mech. Behavior of Materials, Kyoto, Japan, Aug. 1971,
Vol. III, pg. 518.
66
12. L. J. Knox, J. C. Whitwell: Text. Res. J. 41,510 (1971).

13. R. F. Fedors, R. F. Landel: Trans. Soc. Rheol. 9, 195 (1965).
14. S. Kase: J. Polym. Sci., 11, 424 (1953).
15. S. Kaufmann: Faserforsch. u. Textiltechn. 22,406 (1971) and ibid. 463 (1971).
16. K. E. Hansen, W. C. Forsman: J. Polym. Sci., A-2, 7, 1863 (1969).
17. D. A. Stuart, O. L. Anderson: J. Am. Ceramic Soc. 36,416 (1953).
18. H. H. Kausch: J. Polym. Sci., C. (polymer Symposia) 32,1 (1971).
19. A. Tobolsky,H.Eyring: J.Chern. Phys. 11, 125 (1943).
20. I. M. Ward: Mechanical Properties of Solid Polymers, London: Wiley-Interscience 1971,
p.280.
21. A. N. Gent, P. B. Lindley: Proc. Roy. Soc. (London) A249, 195 (1968).
W. K. Ko, Ph. D. Dissertation, California Institute of Technology, 1963.
A. E. Oberth, R. S. Bruenner: Trans. Soc. Rheology 9, 165 (1965).
22. C. K. Lim, Ph. D. Thesis: The Pennsylvania State University, University Park, PA, 1967.
23. L. J. Broutman, S. M. Kirshnakumar, P. D. Mallick: J. Appl. Polym. ScL14, 1477
(1970).
24. R. E. Ely: Polymer Engng. and Sci. 7, 40 (1967).
25. R. A. Dickie, T. L. Smith: J. Polym. Sci. A-2, 7, 687 (1969).
26. J. C. Bauwens: J. Polymer Sci. A-28, 893-901 (1970).
27. R. Raghava, M. Caddell, G. S. Y. Yeh: J. Materials Sci. 8,225-232 (1973).
28. R. G. Dong: Trans. Soc. Rheol. 18, 45 (1974).
29. N. W. Tschoegl: J. Polym. Sci. C 32, 239-267 (1971).
30. T. L. Smith: ibid., 269-282.
31. S. V. Radcliffe: Deformation and Fracture of High Polymers, H. H. Kausch, J. A. Hassell,
R. I. Jaffee, eds. New York: Plenum Press 1973, p. 191-208.
32. J. A. Sauer, K. D. Pae: Colloid a. Polymer Sci. 252,680-695 (1974).
33. W. Schneider, R. Bardenheier: Z. Werkstofftechnik 6,269-280 and 339-348 (1975).
34. W. Retting: Angew. Makromol. Chern. 58/59,133 (1977).
35. A. A. Griffith: Phil. Trans. Roy. Soc. 221,163 (1921).
36. M. L. Williams: Fracture of Solids, D. C. Drucker and J. J. Gilman, eds. New York: Inter-
science, 1963, p. 157.
37. W. G. Knauss: Appl. Mechanics Rev., January 1973, p. 1.
38. M. L. Williams: J. Adhesion 4, 307 (1972).
39. G. P. Anderson, K. 1. DeVries, M. 1. Williams: Int. J. Fracture 9,421 (1973).
40. D. H. Kaelble: J. Appl. Polymer Sci. 18, 1869-1889 (1974).
41. T. L. Smith: Rheology, Vol. 5, F. R. Erich, ed., New York: Academic Press 1969, p. 127.
T. L. Smith, IV, V, and VI Microsymposia on Macromolecules, Main Lectures, London:
Butterworths, 1970, p. 235-253.
42. P. J. Blatz, S. C. Sharda, N. W. Tschoegl: Trans. Soc. Rheol.18, 145 (1974).
43. S. Glasstone, K. J. Laidler, H. Eyring: The Theory of Rate Processes, New York:
McGraw Hil11941.
44. A. Tobolsky, H. Eyring: J. Chern. Phys. 11,125 (1943).
45. S. N. Zhurkov, B. N. Narzullayev: Zhur. Techn. Phys. 23, 1677 (1953).
46. S. N. Zhurkov, E. Yeo Tomashevskii: Zhur. Techn. Phys. 25,66 (1955).
47. S. N. Zhurkov, T. P. Sanfirova: Dokl. Akad. Nauk SSSR 101,237 (1955).
48. F. Bueche: J. Appl. Phys. 26,1133 (191955).
49. F. Bueche: J. Appl. Phys. 28, 784 (1957).
50. F. Bueche: J. Appl. Phys. 29, 1231 (1958).
51. G. M. Bartenev, Yu. S. Zuyev: Strength and Failure of Viscoelastic Materials, Oxford:
Pergamon Press 1968, p. 143fT.
52. E. H. Andrews: Fracture in Polymers, Edinburgh: Oliver and Boyd 1968.
53. V. R. Regel, A. I. Slutsker, E. E. Tomashevskii: Kinetic Nature of Durability of Solids
(in Russian), Isdat. Nauka, Moscow 1974.
54. A. I. Gubanov, A. D. Chevychelov: Fiz. Tverd. Tela 4,928-933 (1962), SOY. Phys.-
Solid State 4, 681-684 (1962).
67
55. G. M. Bartenev: Mekh. Polim.(USSR) 6,458-464 (1970), Polym. Mech. (USA) 6,

393-398 (1970).
56. K. E. Perepelkin: Fiz. Khim. Mekh. Materialov 6, 78-80 (1970).
57. S. I. Mikitishin, Yu. N. Khonitskii, A. N. Tynnyi: Fiz. Khim. Mekh. Materialov 5,
69-74 (1969).
58. A. V. Dobrodumov, A. M. El'yashevich: Fiz. Tverd, Tela 15,1891-1893 (1973).
SOy. Phys. - Solid State, 15, 1259-1260 (1973).
59. e. e. Hsiao: J. Appl. Phys. 30, 1492 (1959).
60. H. H. Kausch, C. C. Hsiao: J. Appl. Phys. 39,4915-4919 (1968).
61. H. H. Kausch: Kolloid-Z. Z. Polymere 236,48-58 (1970).
62. W. Holzmiiller: Kolloid-Z. Z. Polymere 203,7-19 (1965).
63. G. M. Bartenev, I. V. Razumovskaya: Fiz. Khim. Mekh. Materialov 5,60-68 (1969).
64. R. L. Salganik:' Int. J. Fracture Mech. 6, 1-5 (1970).
65. A. N. Orlov, V. A. Petro v, V. I. Vladimirov: phys. stat. sol. (b) 47, 293-303 (1971).
66. B. D. Goikhman: Strength Mater. (USA) 4, 918-922 (1972).
67. Yu. Va. Gotlib, A. V. Dobrodumov, A. M. EI'yashevich, Yu. E. SvetIov: Fiz. Tverd.
Tela 15, 801 (1973), SOy. Phys. - Solid State, 15,555-559 (1973).
68. K. C. Valanis: Report G 378 - ChME - 001 (1974), University of Iowa, Iowa City, USA.
69. F. Bueche, J. C. Halpin: J. Appl. Phys. 35, 36-41 (1964)
70. L. Holliday, W. A. Holmes-Walker, J. Appl. Polym. Sci. 16, 139-155 (1972).
71. P. I. Vincent: Polymer 13,557-560 (1972).
72. D. C. Prevorsek: J. Polym. Sci. C32, 343-375 (1971).
73. V. M. Poludennaja, N. S. Volkova, D. N. Archangel'skij, A. G. Zigockij, A. A. Konkin:
Chim. Volokna 5, 6-7 (1969).
74. H. Kriissig: Textilveredelung 4, 26-37 (1969).
75. V. R. Regel, A. M. 'Leksovsky: Int. J. Fracture Mechanics 3,99 (1967).
76. C. B. Henderson, P. H. Graham, C. N. Robinson: Int. J. Fracture Mechanics 6,33-40
(1970).
77. G. M. Bartenev, l. V. Razumovskaya: Fiz. Tverd. Tela 6,657-661 (1964), SOy. Phys.
- Solid State 6, 513-516 (1964).
78. H. H. Kausch: J. Polym. scL C32, 1-44 (1971).
79. S. I. Mikitishin, A. N. Tynnyi: Soviet materials seL 4/3, 187-188 (1968).
80. S. S. Sternstein, J. Rosenthal: Toughness & Brittleness of Plastics, Adv. Chemistry Series,
A.e.S., Washington, D. e. (1974).
81. W. Retting: Orientation in Polymers Programme, Polystyrene, Rep. Results of BASF,
2nd part, March 1972, Macromol. Division, I. U.P.A.e.
82. I. Narisawa, T. Kondo: J. Polymer Sci. - Polymer Phys. Ed. -11, 1235-1245 (1973).
83. P. Stockmayer: Stuttgarter Kunststoff-Kolloquium 1977, Kunststoffe 67,470 (1977).
84. H. H. Kausch: Materialpriifung.
85. J. W. S. Hearle: J. Textile Inst. 68/4, 155-157 (1977).
86. A. V. Savitskii, V. A. Mal'chevski, T. P. Sanfirova, L. P. Zoshin: Polymer Sci. USSR 16,
2470-2477 (1974), Vysokomol. Soyed. A 16,2130-2135 (1974).
68
Chapter 4
Strength of Primary Bonds
I. Covalent Bonds . 69
A. Atomic Orbitals 69
B. Hybridization . 71
C. Molecular Orbitals 72
D. Multiple Bonds . 73
II. Bond Energies 73
A. Electronic Energy and Heat of Formation 73
B. Binding Energy of Excited States and Radicalized Chains 80
1. Electronic Excitation 81
2. Ionization 82
3. Radicalization 84
III. Form of Binding Potential. 86
I. Covalent Bonds
In the last chapter, in the section on molecular theories of fracture, almost through-
out an Arrhenius equation has been used to describe the activation of element break-
age. The energy of activation, Uo , frequently turned out to be equal to (or was
assumed to be equal to) the dissociation energy of the weakest main chain bond.
Before further analyzing the kinetics of element - and possibly chain - breakage
a definition of the mechanical strength of a bond and of a chain have to be given.
In order to do this basic results of quantum chemistry [1, 2] are recalled in this
chapter which concern the "strengths" of intramolecular bonds and factors in-
fluencing the binding potential such as electronic excitation or ionization.
A. Atomic Orbitals
Covalent chemical bonds between atoms of the same or a different species rely on
the interaction of the outermost - or valence - electrons. Even though one speaks
of electrons one should rather think of electron clouds, i.e. of electronic density
distributions. The radial and angular distribution of the electron density is described
by "one electron" wave functions \}I - also called atomic orbitals - which are
derived as a solution of the quantum mechanical Schrodinger equation:
69
4. Strength of Primary Bonds
(4.1)
Here h is Planck's constant, mr the reduced mass of an electron with charge e,

V the Nabla operator, and E the energy eigenvalue of an electron moving within a
potential V. The term in brackets constitutes the Hamiltonian operator H.
The Schrodinger equation has a solution only for certain discrete energy values
E. Different values of E mean different forms of the electronic density distributions
which are characterized by the principal quantum number n, the angular quantum
number Q and the magnetic quantum number m.
For the simplest atomic system, the hydrogen atom, which consists of one pro-
ton and one electron the potential V is _e 2 /f. In this case energy levels are obtained,
which are degenerate, i.e. the energy levels for different I and m coincide. The cor-
responding wave functions, however, do depend on the three quantum numbers n, I
and m. It should be noted that the radial distribution, R(r), and the angular distribu-
tion 8(8, (1) may be separated:
( 4.2)
The quantum numbers are subject to the condition that
Q = 0,1,2 ... n-l

m = 0, ±1, ±2 ... ±Q.
It is customary to denote the atomic orbitals by the spectroscopical terms s,

p, d ... if Qis 0, 1, 2 ... respectively.
The first few wave functions for a hydrogen-like atom with a charge of +Ze of
the nucleus then have the following form:
Is RlO(r) = (Z/ao)3/2 2 exp (-Zr/ao) (4.3)

2s R20(r) = (Z/ao)3/2 (2 - Zr/ao) (exp - Zr/2ao) (1/2 0) (4.4)
2p R21 (r) = (Z/ao)3/2 (Zr/2ao v'6j exp (- Zr/2ao) (4.5)
and
s 8 00 (8, (1) = 1/2 y'rr (4.6)

px 8Ix (8, (1) =(.../3/2 y'rr) cos 11 sin 8 (4.7a)
py 8 ly (8, (1) =(v'3/2 y'rr) sin 11 sin 8 (4.7b)
pz 81z (8, (1) = (v'3/2 y'rr) cos 8. (4.7c)
Here ao is the radius of the first Bohr orbit in hydrogen (0.529 . 10- 10 m).
According to the Pauli exclusion principle every atomic orbital may be occupied
by one or two electrons which in the latter case have to have their spins aligned
antiparallel to each other (paired electron spins).
For many-electron atoms an exact analytical solution of the Schrodinger
equation (Eq. 4.1) is not possible. Nevertheless atomic orbitals may be obtained
70
I. Covalent Bonds
by an iterative procedure, which starts by placing the appropriate number of elec-

trons on hydrogen-like one-electron orbitals. With the potential obtained from such
a charge distribution a better approxiination of the first orbital can be calculated
which in tum is considered in the recalculation of the second orbital and so on until
no further corrections become necessary. Again these orbitals are denoted by 1s,
2s, 2p, 3d if the quantum numbers n and Qare 1 and 0, 2 and 0, 2 and 1, 3 and 2
and so on.
In a many-electron atom - as for instance U - one finds a particular electron
in any of the accessible orbitals (1 s, 2s) and in any of the two possible states of
spin (ms = + 1/2 or -1/2). Since electrons are indistinguishable the energy of the
atom must not be affected by interchanging the coordinates of any pair of elec-
trons. The wave function belonging to one eigenvalue E will have to encompass,
therefore, all terms resulting from an interchange of electron numbering.
In the case of U the ground state has the wave function:
ls a (1) lsj3 (1) 2s a (1)

'1'= (4.8a)
2s a (3)
or
'I' = Is a(I) Isj3 (2)2s a (3) - Is a (2) Isj3 (1) 2s a (3) + ... (4.8b)
Each of the six terms of the fully written determinant (4 .8b) contains an orbital
and a spin function for each of the three electrons. The spin function a (I) may
denote that the electron 1 has a spin quantum number ms = +1/2,13(1) corresponds
to m = -1/2.
B. Hybridization
It can easily be shown that an atom with one electron in each of the 2s-, 2px-,
2py-, and 2pz-orbitals has a total charge density distribution which is completely
spherical, i.e. independent of e and cp. The s- and p-atomic orbitals are not the only
functions that lead to a spherical distribution of the total charge. A large number
of suitable linear combinations of the s- and p-orbitals do have the same property.
These linear combinations are called hybrid orbitals or hybrids. Depending on the
number and kind of atomic orbitals used in hybridization the hybrids are denoted
as sp, Sp2, sp3, dsp etc. The four normalized and orthogonal Sp3 -hybrids (of a
C-atom) have the form:
(4.9a)
(4.9b)
(4.9c)
71
(4.9d)
The individual hybrid is, of course, non-spherical, i.e. the electron density dis-
tribution has a characteristic shape. In this particular case the directions of maximum
electron density point into the corners of a regular tetrahedron.
The electronic energy of a C-atom with electrons in hybrid orbitals is higher
than that of the ground state. In carbon the ground state will be formed by two elec-
trons in each of the lowest orbitals: 1S2 2S2 2p2. Hybridization involves the change
from pyramidal orbitals (p3) to tetragonal ones (Sp3); it requires, that three p-or-
bitals be occupied. The first step is, therefore, that one electron be promoted from
2s to 2p which necessitates an energy of promotion. The effect of promotion and
hybridization is, however, that four rather than two electrons are available for bond-
ing. The gain in binding energy of the two additional C-H bonds (90 to 100 kcal/mol
per C-H bond) more than balances the energy expended for promotion and hybrid-
ization.
c. Molecular Orbitals
Within the limited scope of this book a mere terminology of molecular orbital theory
can be supplied - intended only for the discussion of bond energies. Rigorous quan-
tum mechanical calculations for any molecule being more complex than the hydrogen
molecule are not feasible. Therefore, approximate methods for calculation of the
orbitals of the bonding electrons have been developed [1,2]. The two most impor-
tant ones are the molecular orbital (MO) method relating to the work of Hund, Mulli-
ken, and Hiickel [la] and the bond orbital method, the Heitler-London-Slater-
Pauling (HSLP) method [1 b]. In the MO approach electrons are placed in one-
electron-wave functions which belong to the molecule as a whole with each nucleus
in its proper configuration. In its crudest form these wave functions are obtained as
a linear combination of the atomic orbitals (LCAO) of all atoms forming the molecule
[2a].
In bond orbital theory the wave functions of the molecule are derived from the
wave functions associated with the different bonds of a molecule, Le. from the bond
orbitals. In the LCBO model the molecular orbitals are a linear combination of bond
orbitals which in turn are a combination of the atomic orbitals or hybrids forming
the bond in question [2b]. In the HLSP-method the many-electron wave functions
are a sum of product functions which contain a HeitIer and London type of factor
(space function and spin function) for each bond of the molecule.
Bond orbital and molecular orbital approach lead to the common conclusion
that bonding will occur only at directions where the atomic orbitals have large values
and only if the atoms are close enough to each other so that the atomic orbitals
overlap, i.e. if the "electron clouds" partly penetrate each other.
A covalent bond is associated with an electron pair and hence subject to the
quantum mechanical phenomenon of electron exchange: in a molecule A-B the
electron from A may also be found near the nucleus of atom B and vice versa. It is
72
II. Bond Energies
this additional exchange energy term which significantly deepens the interatomic
potential and leads to a distinct potential minimum - prerequisite for a stable
binding. Thus a bond can be formed if a bonding molecular orbital exists and if
it is occupied by an electron or an electron pair. In the ground energy state of a
molecule paired electrons occupy the molecular orbitals with the lowest energy.
D. Multiple Bonds
So far multiple bonds between atoms of the same or a different species such as carbon,
nitrogen, and oxygen have not been explicitly discussed). At this point attention
should be drawn to the fact that the two (or three) molecular orbitals involved in
forming a double (or triple) bond are usually not of the same kind but of different
character. In ethylene(H 2 C=CH 2) all six atoms lie in one plane. Therefore, one
may take three sp2-hybrid orbitals of each C·atom to form the four C-H bonds
and the one axially symmetric C-C bond, the a bond.
The two remaining pz orbitals are combined into the so called (molecular) 1f
bond orbital. Since the pz orbitals are directed perpendicular to the 1f bond which
they are forming their overlap is smaller than that of orbitals forming a a bond.
Also the regions of maximum overlap are not in the xy plane between the C-atoms
but rather above and below this plane. Because of this orientation of the pz orbitals
the interaction between conjugated 1f bonds is so strong that in fact it leads to a
delocalization of the 1f electrons. In view of the geometry of the 1f bond it is easily
realized that rotation of a group around multiple bonds is forbidden [Ic].
II. Bond Energies
A. Electronic Energy and Heat of Formation
The term bond energy requires a definition. The potential energy·scheme of an ar·
bitrary bond A-B in a poly atomic molecule given in Figure 4.1 serves to illustrate
this point. As pointed out previously one has to take into account that in poly·
electronic atoms the valency state of the atom may lie above the respective ground
state. If this is the case no binding can occur between two atoms in their ground
states; if they encounter each other their potential energy will rise. At some inter-
atomal distance the potential energy of the system will approach the energy of the
atoms in their valency states (Fig. 4.1, dashed line) and a transition to the binding
state can occur. The intrinsic binding energy (wahre Bindungsenergie) E is, therefore,
the difference between the energy of the molecular ground state and that of the
valency states at infinite interatomal distance. The energy of dissociation D is smaller
than E, namely by the zero point energy (Nullpunktsenergie) h v/2 of vibration
and by the sum P of energies of promotion, hybridization, and polar and steric
rearrangements necessary to attain the valency state. The difference between zero
point energy and the maximum of the potential energy curve is the activation
73
'.
interatomal distance
Fig. 4.1. Scheme of the potential energy of an arbitrary bond in a polyatomic molecule.
D energy of dissociation, E intrinsic binding energy, P energies of promotion, hybridization
and of polar and steric rearrangements, Uo activation energy of bond dissociation, ro equilibrium
bond length, hv/2 zero point energy of vibration.
energy Do of bond dissociation. In diatomic molecules Do and D generally have the

same value and represent the bond energy.
In molecules with three or more atoms the assignment of a specific increment
of energy to a particular bond is complicated by the presence of the above mentioned
energy term P for promotion and electron reorganization, by the occurrence of delo-
calized electrons, and by the fact that the dissociation energies of successively broken
identical bonds are different. The latter statement may be illustrated by the disinte-
gration of CH 4 .1t must be assumed that the four hydrogen atoms in CH 4 are equiv-
alent, all C-H bonds are formed by hydrogen Is orbitals and carbon 2Sp3 orbitals.
Abstraction of the first hydrogen atom requires an energy of 425.8 kl Imol
(101.7 kcal/mol) and changes the methane atom into the methyl radical CH 3 . From
symmetry considerations confirmed by ESR investigations it has been concluded
[4] that the methyl radical is planar, with the unpaired electron undoubtedly in a
2p orbital of the C-atom. The bond character of the C-H bonds remaining in what
now is the methyl radical has changed, therefore, from the tetrahedral configuration
of sp3 hybrid orbitals into the planar one of Sp2 hybrids. The dissociation energy
of one of these C-H bonds is 477.3 kllmol (114 kcal/mol). The third hydrogen
has a dissociation energy of 418.7 (100 kcal/mol), the fourth of 341.2 kllmol
(81.5 kcal/mol). The total energy gained in forming CH 4 from its constituent atoms
is, therefore, 1663 kllmol (397.2 kcal/mol), the average C-H bond energy
415.7 kJ/mol (99.3 kcal/mol) [5]. It is this average derived from the energy offor-
mation of the molecule which is usually termed bond energy (mittlere Bindungs-
energie).
The bond energies of saturated and unsaturated homologues of hydrocarbons
may be treated in some detail since they will give us some understanding of the cor-
relation of bond energies in poiyoiefins. In saturated hydrocarbons with localized
electrons it is reasonable to equate the energy of a bond with the energy of the
74
II. Bond Energies
electrons forming that particular bond. Theoretically the electronic energy of a

molecule may be calculated using any of the mentioned approximations, for in-
stance the LCBO method [2b, 6]. In that case the molecular orbitals for the 2N u-
electrons of a hydrocarbon with N bonds may be written as a linear combination of
N bond orbitals cPj:
N
Wj = ~ Cij CPj. (4.10)
i=1
One assumes that the wave functions Wj are a solution of the equation
(4.11)
or
(H-E) Wj = O. ( 4.12)
If one defines the notations
Hij =f CPi H CPj dv (resonance integral) (4.13a)

Sjj =f cPj CPj dv (overlap integral), (4.13b)
introduces Eq. (4.10) into Eq. (4.12), and multiplies that equation from the left
hand side by CPj one obtains
N
.~ Cji(Hji-ESji )=0 j=1,2, ... N. (4.14)
1=1
The system (4.14) of equations has a nontrivial solution only if the determinant
of the quantities in brackets is zero. Since in principle the coefficients Hij and Sij
are derivable from the electron wave functions the secular determinant of the sys-
tem (4.14) may be solved to yield the energy eigenvalues El to EN. The total
electronic energy E of the molecule then is
N
E = ~ 2 Ej . (4.15)
j= 1
Following the treatment of Brown [6] it is assumed that the Coulomb integral,
Hmm, of an orbital CPm depends only on the type of bond in which it resides, thus
all C-H bond orbitals will be assigned the same value a of the Coulomb integral.
The term is one half of the binding energy of an isolated C-H bond. The resonance
integral, Hmn, will be assumed to be zero unless m and n denote adjacent bonds. In
that case Hmn still depends on the type of neighboring bonds. In Table 4.1 the as-
signments of Brown for the bond interaction parameters in saturated hydrocarbons
are listed.
75
Table 4.1. Bond Interaction Parameters
Coulomb parameter Resonance parameter Overlap integrals

Hmm Hmn Smn
Bonds Parameter Bonds Parameter Bonds Parameter
C-H C-H, C-H (3 C-H, C-H S

C-C C-H, C-C 8(3 C-H,C-C 8S
C-C, C-C n(3 C-C, C-C nS
Table 4.2. Electronic Energies of Paraffins
Paraffin Electronic Energy E Etheor Eexp+W

(S), = 6.5)
kllmol kllmol
Methane 80< - 24S), -1765 -1665

Ethane 140< + 2h),-(24 + 2402)S)' -2923 -2828
Propane 200< + 4h),-(28 + 400 2 + 4n2)S)' -4088 -4001
Butane 260< + 6h),-(32 + 560 2 + 8n2)S)' -5253 -5175
isoButane 260<+ 6h)'-(36 + 480 2 + 12 n2)S)' -5260 -5182
Pentane 320< + 8h),-( 36 + 720 2 + 12n 2 )S), -6418 -6350
isoPentane 320<+ 8h)'-( 40 + 640 2 + 16n 2 )S)' -6425 -6358
neoPentane 320< + 8h),-(48 + 480 2 + 24n 2 )S)' -6438 -6370
Hexane 380< + lOh),-(40 + 880 2 + 16n 2 )S), -7583 -7525
2-Methylpentane 380< + lOh),-(44 + 800 2 + 20r?)S), -7590 -7532
3-Methylpentane 380< + 10h),-(44 + 800 2 + 2On2)S), -7590 -7529
2,3-Dimethylbutane 380< + 10h),-(48 + 720 2 + 24n2)S)' -7597 -7535
2,2-Dimethylbutane 380< + 10h),-(52 + 640 2 + 280 2 )S), -7603 -7543
In Table 4.2 and 4.3 the electronic energies of a few hydrocarbons and of their
radicals are given in terms of Brown's parameters. It will be noted that to a zeroth
approximation (S = 0) the total electronic energy E of a molecule or a radical is given
by the sum of the energie& of the corresponding number of isolated C-C and C- H-
bonds. In the first approximation also the energetic interactions of neighboring
bonds are considered (terms to the first power of S as listed in Tables 4.2 and 4.3)
and higher powers of S are neglected. Such terms would be necessary, however,
to account for the effect of third neighbors within the molecule. That effect is
clearly revealed by analysis of thermochemical data [7].
If it were possible to determine the numerical values of the bond interaction
parameters by simply using molecular geometry and atomic constants one would
have a direct and independent way to calculate bond energies. So far, however, it
has proven very cumbersome to derive the explicit values of the bond energy para-
meters 0:, h, 'Y, 8 2 , T/ 2 , and S directly and sufficiently correctly from molecular
wave functions. Usually thermochemical data are employed to obtain the electronic
energy of a molecule. For example the electronic energy E of a hydrocarbon can be
76
Table 4.3. Energy* of Hydrocarbon Radicals R
Radical R Electronic Energy E(R-H)-E(R) D(R-H)* AHf(298 K)*

kJ/mol kJ/mol
Methyl CH3 601-12S)' Fl + 3F 2 436 134
Ethyl H3C-CH2 1201 + 2h),-(16 + 200 2 )S)' F\ + 2F2 414 110
n-Propyl H3C-CH2 -CH2 1801 + 4h),-(20 + 360 2 + 4r)2)S)' F\ + 2F2 411 86
iso-Propyl H3C-CH-CH3 1801 + 4h),-(24 + 320 2 + 4r)2)S)' F\ + F2 394 72
n-Butyl H 3C-(CH 2 )2 -CH 2 24a + 6h),-(24 + 520 2 + 8r)2)S)' F\ + 2F2 411 67
iso-Butyl H 3 C-CH(CH 3 )-CH 2 24a + 6h),-(28 + 440 2 + 12r)2)S)' F\ + 2F2 411 57
sec.-Butyl H3C-CH2-CH-CH3 240: + 6h),-(28 + 480 2 + 8r)2)S)' F j + F2 394 50
tert.-Butyl (CH 3 hC 2401 + 6h),-(36 + 360 2 + 12r)2)S), Fj 381 28
* Energy values from Refs. [8a] and [9 J. -

I :!j
0
::l
p..
t'l"l
::l
(1l
....
-..J o,s.
-..J ~
calculated from the enthalpies Ha of atomization of hydrogen, L of sublimation

of graphite, and ~Hf of formation of the hydrocarbon from H2 and Cn :
E(C n H2n + 2) = - n L(C) - (n + 1)Ha + ~Hf - w. (4.16)
The term W accounts for the fact that the minimum of a binding potential is by
the zero point vibrational energy (I/2 hv) lower than the corresponding dissociation
energy (Fig. 4.1). Since the relevant frequencies are of the order of 2.5 . 10 13 s-l
(C-C skeleton) and 9 . 1013 s-l (C-H) the zero point vibrational energies range
between 5 kJ/mol and 18 kJ/mol.
A quantitative analysis of the quantum mechanical calculations through ther-
mochemical data will be attempted, although considerable uncertainties exist. These
are caused by the limitations of the used first order approximation, the lack of con-
sideration of steric interaction of atoms, and of the difficulties to determine W. The
analysis is firstly based on the fact that the differences between the electronic
energies of the listed hydrocarbons and their radicals can be expressed by only four
quantities F j :
Fl =-2a+ 1282 Sr (4.17)

F2 = 4(1-8 2) Sr (4.18)
F3 = -4(1 - 282 + 7]2) Sr (4.19)
F4 = -2a- 2hr + 247]2 Sr. (4.20)
Brown [6] determined F2 to be 15.5 kJ/mol (from the radical dissociation

energies listed in Table 4.3) and F 3 to be -6.7 kJ Imol (from the energy differences
found for 5 pairs of isomers listed in Table 4.3).
Since these difference values are not so much affected by the systematic un-
certainties mentioned above they will be adopted here along with the value of
300 kJ Imol for F 4; F 4 is the electronic energy of a C-C bond interacting solely
with C-C bonds as in (CH3hC-C(CH3h- Further the standard values (at
298.15 K and 1 bar) L = 717.2 kJ/mol and Ha = 436.2 kJ/mol [7] and a ~ Hf (CH 4) =
= -75 kJ/mol have been used [6]. The zero point vibrational energy of the 9 vibra-
tional modes of the -C- H and H-C-H bonds of methane is estimated at 100 kJ Imol.
The total electronic energy of methane then is derived from Eq. (4.16) as E (CH 4)
= -1765 kJ /mol. From Eqs. (4.17) and (4.18) it follows that 6.2 < Sr < 8.4 kJ /mol.
Employing E (CH 4), F 2 , F 3 , and F4 as given above and a value of
Sr = 6.5 kJ Imol one obtains

a = -200 kJ/mol
hr = 55 kJ/mol
8 2 = 0.4 kJ/mol
7]2 = 0.06 kJ/mol.
On the basis of these values the electronic energies of the hydrocarbons have
been determined and listed as Etheor in Table 4.2. For comparison the thermo-
chemically determined enthalpies of formation are referenced as Eexp + W. The
latter values are consistently larger than Etheor although the difference decreases
rather than to increase with the number of vibrational modes available to a molecule.
78
II. Bond Energies
Table 4.4. Dissociation Energies of Primary Bonds in Organic Molecules and Polymer
Chains *
H- 431 406 394 373 436 507 427 360 469 365 427 347 300
F- 453 444 440 427 245 524 377 448
Cl- 352 339 339 331 360 205 344 419 285 314
OH- 381 381 385 381 427 419 469 356
O=CH- 314 297
NH,- 331 3Z7 323 323 411 419
N=C- 461 545 451
Q9
= = "'V=
'" ~~ ,.-,,....,.....,
=== '" '"
~= = ~-
~ 'I '", Lf
9=9
~
L..:l
L) ~
'" '"
3'I =
~
L..:lL.JU
'1/
~
3, ~
~
'"I
~
~
VI 6 ~ '"
L..:l '"I '"
u-u 6'"
I I I I I I I I I do
CH)
OCH-
276 355 -CH 1-R
,CH)
1 302
OCH-
_ 260 230 293 264 210 323 32. -C-R
'CH)
0
Oc- 377 R-CHryH-R'
0 427 381 348 327 423 499 423 335 3.5 -O-R
CN-CH,- 306 190 311 -CH1O-R

H,
o·C- 314 297 251 336 -(CH 1i1 O-
CH hC_O_ -CH-R
379* 3ll I
0'" OH
CH hc _ -CH-R
344 323 318 I
0'" CH)
CH)-O- 335 335 339 3Z7 316 -CHCI-R
CH=C- 419 457 432 JV -CHF-R

-CH-R
CHrCH- 377 377 356 198 I
CN
CH),
CH)-C- 335 314 03 377 30l -C 6 H,-R
CH/
CH hCH _ -CH-R
348 270
CH{ C6HS
CH)-CH 1- Ill, RNH-C6HC NH-
~
CH)- lSI, -~~R
6HS
R~-c,H4~-O-CHl-
172
0 0
185 -(CH 1kCO-R
106 -{H 1-C6HC R

I
~=c:>
I
=
'"
~
'"
= '"
~- ~
I I
~
~;r
I I
I I
'"
=
A-,\",
~== ~ ~
'" '" G~ '"

~ '" '"
~ ;r
~
.I
~ ~ ~
~ ci" = ci" ci" ci"
* Upper and left hand side: after van Krevelen [91

Lower and right hand side: after Stepan'yan et al. [101
79
The treatment may be extended to deal with the energy of alkyl radicals if one
assumes that the single unpaired elec.L on on the "radicalized" carbon atom occupies
of 2P1T molecular orbital and does not change the interaction of other electrons on
other orbitals. As mentioned earlier this assumption is in accord with ESR-investi-
gations [4]. In electronic terms the dissociation eq.ergy D of a bond Rl - R2 can
then be formulated as
(4.21)
In Table 4.3 the electronic energy of hydrocarbon radicals and R-H dissociation
energies are listed together with thermo chemically determined values. The difference
values of electronic and thermochemical data are in correspondence with each other.
The absolute value of F 1 = D(R-H) + 1/2 hv(R-H) can be calculated from Eq.
(4.17) to be 431 kllmol; since D(R-H) for tertiary carbon atoms is 381 kllmol a
contribution of 50 kllmol must be assigned to the zero point vibrations. This con-
tribution is certainly too high by ~30 kllmol. The discrepancies between cal-
culated and measured dissociation energies of C-C bonds, however, lie between 0
and 6 kllmol for the bonds covered by Tables 4.2 and 4.3. The experimentally
determined dissociation energies of small molecular groups are listed in the upper
and left hand side of Table 4.4 [after van Krevelen, 9]. For these molecular frag-
ments the activation energy of bond rupture and the dissociation energy are assumed
to correspond. In the lower right hand side we have listed the dissociation energies
obtained for Sp2 and sp3 bonds by Stepan'yan et al. [10]. In their semi-empirical
approach the authors calculate the bond dissociation energies RI-R2 byextrapo-
lation from those of the series of compounds R-X, X representing F, H, CI, Br,
and I.ludging the available experimental and theoretical determinations of bond
energies of chain molecules it must be stated that still considerable differences with
respect to the depth of the binding potentials for main chain bonds exist.
Before leaving the subject of bond energies one should briefly discuss the mean-
ing of the term bond strength. In quantum chemistry the term bond strength, or
more explicitly the bonding strength of atomic orbitals, refers to the excentricity
of a (hybrid or nonhybrid atomic) orbital and is measured by the maximum
angular intensity of the wave function as compared with the spherically symmetrical
s-orbital [1]. So the strength of an sp3 hybrid is 'Ir 1(8 =54° 44'; <p =45°)1 (;1T) 1/2 = 2.
Other hybrids of interest have the following bond strengths: the (pyramidal) p3
hybrid 1.732, the (plane triangular) Sp2 hybrid 1.991, and the (linear) sp hybrid
1.932.
Throughout our further discussions, however, the term bond strength will have
a more technological meaning, i.e. that of the force required to break the bond
(cf. Chapter 5).
B. Binding Energy of Excited States and Radicalized Chains
In as much as the covalent binding force between any two atoms rests on the electron
density distribution between (and around) these atoms any change of the electron
80
II. Bond Energies
distribution necessarily affects the bond strength. Since the states of different elec-
tronic energy of a molecule are states of different electron distribution they are also
states of different interatomal forces.
This is also true, of course, for molecular ions, i.e. for molecules where one or
more electrons have been removed completely. As already seen for methane the
character of a bond and its energy are also affected through the radicalization of the
atom group to which it belongs.
1. Electronic Excitation
The electronic excitation of a polymeric network chain can be caused by elec-

tromagnetic (light, UV, '}') or particle irradiation. The transfer of energy from the
impinging particle or quantum of radiation to the electron requires that sufficient
energy is available to reach an excited state and that an interaction mechanism exists.
In the case of visible light a photon of, say, a wavelength of 330 nm has enough
energy to break a C-C bond. However, the photon will not be absorbed by alkanes
and there are no electronic states of that or a smaller energy which could be ex-
cited. To be effective a photon has to be absorbed and to interact with a binding
electron. This interaction occurs either directly or indirectly through the mechanisms
of energy transfer by exciton diffusion, single step transfer or absorption of fluores-
cence light emitted by the same or another (impurity) molecule [11]. The nature
and sequence of these important processes which determine the photochemical
stability - or instability - of polymers will not be treated here in detail. One is
interested, however, to know the levels of energy where the excitation of electrons
- or the ionization of molecules - begins and to know the changes of binding energy
caused by excitation - or ionization - respectively.
kcal/mol , - - - - - - - - - - - - - - , 10
Ele.v)
200
- -- - --- - -----
-:...;;-:.;;,--.-
150
~
01
Q;
; 3P + 10
c
~
.!! 100
0
a.
50
o o
Fig. 4.2. Interatomal energy of ex-
mteratomat distance cited states of a C2 molecule 112 \.
81
The potential curves of different excited states of the C2 -molecule illustrate the
effect of electronic excitation on bond energies (Fig. 4.2). We would like to point
out that the dissociation energies of the different molecular states shown range from
350 to 119 kJ/mol (84 to 28 kcal/mol). These dissociation energies are generally
smaller than the energy differences caused by the electronic excitation. The excited
states of the (many-electron) C-atoms are denoted in the usual manner as S, P, D.
Since two excited atoms may combine in a number of ways there are more than one
molecular state corresponding to any two atomic states.
Transitions from the ground state of a molecule to an excited state usually occur
from a bonding orbital (a or 1r electron) or a non-binding orbital (n electron) to an
antibonding a* or 1r* orbital. In Table 4.5 are listed the electronic transitions for
some common high polymers as given by [13]. If an electronic excitation of a net-
work chain occurs, it leads to decreased bond energies. This will reduce the mechanical
stability of a loaded polymeric network and thus contribute to chain scission, frac-
ture initiation or facilitated crack propagation. It is readily seen from Table 4.5,
Table 4.5. Electronic Transitions for some Polymersa
Polymer Amax' nm hvmax Transition

kllmol kcal/mol
Polyethylene <150 >802 >191.5 (J -+ u*

Polybutadiene 180 670 160 1f -+ 11'*
Polystyrene 187-260 461-645 110-154 b
Poly(methyl methacrylate) 214 561 134 n -+ 1T*
Poly(ethylene terephthaiate) 290 415 99 n -+ 1T*
Poly(ethylene terephthalate) 240 502 120 b
a As solid films (after Brenner and Kapfer, 112».

b 1T-electron system involved.
however, that the energy values of the lowest lying excited states are larger than
that of any C-C bond. One does not expect, therefore, that the recombination of
any chain radicals will lead to energy quanta which are large enough to cause elec-
tronic excitation of the reacting radicals or of neighboring molecules.
2. Ionization
The complete removal of an electron from a molecule, Le. the ionization, requires
more energy than electron excitation and has an even stronger effect on the dissocia-
tion energy. In Table 4.6 the dissociation and ionization potentials of the first few
hydrocarbons have been collected from various sources. It is assumed that ioniza-
tion is first accomplished through removal of the least bound electron from the
highest occupied orbital and also that this orbital embraces the whole molecule.
82
Table 4.6. Dissociation and Ionization Energies of Hydrocarbons
Molecule (R 1-R 2 ) D(R 1 -R 2 )a I(R 1 -R 2 )b Radical Ion I(R 2 )C I(RI-R 2 ) - I(R 2 ) D(R1-Ri)
kllmol eV kllmol (Ri) eV kllmol eV kllmol kJ/mol
Hydrogen H2 436 15.43 1490 H+ 13.6 1314 1.83 176 260

Ethylene C2 H6 352 11.76 1134 CH+ 9.82 950 1.94 184 167
3+
414 11.76 1134 C2 HS ,;;;; 8.4 ,;;;; 812 ::'3.36 326 88
Propylene C3HS 347 11.21 1084 CH+ 9.82 950 1.39 134 213
3+
350 11.21 1084 C2H s + ,;;;; 8.4 ,;;;; 612 ::.2.81 ::.272 ,;;;; 75
406 11.21 1084 nC 3H7 < 8.1 < 783 >3.11 >301 <109
393 11.21 1084 isoC3H~ < 7.5 < 724 >3.71 >360 < 33
n Butane C4 H10 347 10.8 1042 CH+ 9.82 950 0.98 92 255
3+
347 10.8 1042 C2H S ,;;;; 8.4 ,;;;; 812 ::.2.4 ::.230 ,;;;;117
347 10.8 1042 C 3H; < 8.1 < 783 >2.7 >260 < 88
a From Table 3.4.

b From Landolt Bornstein 1/3,6. Autlage, Springer Verlag, 'Berlin 1951, p. 363.
C From F. A. Elder, C. Giese, B. Steiner, M. Inghram, J. Chern. Phys. 36, 3292 (1962).
.....
.....
t::tI
o
S.
tT:I
00
w i·
The dissociation energies D(R J - R;) in the last column of Table 4.6 can be cal-
culated from the ionization energies according to the relation
(4.22)
The values derived in that (indirect) way are given in parenthesis because of the
ambiguities that are still involved in obtaining the ionization potentials from elec-
tron impact or photo ionization efficiency measurements. For the molecules listed
here it is clearly apparent that the dissociation energies of the positive molecular
ions are much smaller than those of the non-ionized molecule. (It should be noted
that the dissociation energies of excited states are not necessarily always smaller
than those of the corresponding ground states. But in the organic compounds here of
interest dissociation energies are reduced by electron removal.)
It has been emphasized here that the ionization should lead to a positive molec-
ular or radical ion because the electronic energy of negative ions may be quite dif-
ferent. These differences are not expected and not explained in molecular orbital
theory. If no electron repulsion term is explicitely introduced then the same amount
of energy will be predicted for the removal of an electron and for placing an electron
onto such an orbital. These differences can be quite considerable. For H2 the state
H+ + H- lies above the state H + H by an amount of 12.8 eV (1230 kllmol). The
energy difference between Ri + R2 and R J + R2 is given by I(Rd -E' (R 2 ) where
E' (R 2) is the electron affinity of R2 (i.e. the energy which is released when the
atom or radical R2 combines with an electron to form a negative ion). As is illustrated
by Table 4.7 the electron affinities of saturated hydrocarbons are of the order of
1 eV (96 kllmol) or less, the differences between I(R) and E(R) are of the order
of 7 eV (672 kllmol) or more. The dissociation of a hydrocarbon molecule R J - R2
into the two radical ions Rl and R; generally requires an amount of 672 kllmol
or more in addition to the energy required for homolytic scission into the radicals
R J and R 2 . Homolytic chain scission, therefore, is a much more probable mode of
molecular disintegration (of saturated hydrocarbons) than ionic dissociation.
3. Radicalization
The removal of a hydrogen atom from a hydrocarbon leads to the formation of a

radical. Its electron distribution differs from the original one near the radical
Table 4.7. Electron Affinity and Ionization Potential
Radical Electron Affinity * Ionization potential I(R) - E(R)

R E'(R) leV] I(R) leV] eV kl/mol
H 0.80 13.60 12.8 1230

CH 3 1.08 9.82 8.74 840
C2 HS 0.89 8.4 7.5 720
n-C 3H7 0.69 8.1 7.4 710
* Values of E(R) taken from Handbook of Chemistry and Physics, 53 rd Edition, CRC Press,
Cleveland 1972/73, p E 55.
84
II. Bond Energies
C atom. In saturated hydrocarbons the 4 Sp3 hybrid orbitals change to 3 Sp2 hy-
brids. The radical free electron is considered to be located in a completely separated
1f eigenfunction which has no interaction with the a-system. The strength of a bond
connecting a C atom with the radicalized C atom, t, is now mainly determined by the
interaction of an Sp2 with an sp3 hybrid. The energy of a bond formed by an over-
lap of an Sp3 hybrid and an Sp2 hybrid is not smaller than that formed by two Sp3
hybrids. This is concluded from the dissociation energies of the C-C bonds in
RCH 2 -CH2 R (337 kJjmol) and RCH 2'--C 6 H s (381 kJjmol). The assumption that
replacing sp3 orbitals by sp2 orbitals in bonds will not reduce the bond energy per
se is further supported by the observation that the bond energy of the C(sp2)-H
bond in H2 C=CH 2 is the same as the value of the C(sp3)-H bond in H3C-CH3
(436 kJjmol). The increased energy of dissociation of the C(sp2)-H bond in the
methyl radical (477 kJjmol) leads to a similar conclusion.
The C-C bonds in Q position to the carbon carrying a free electron are, there-
fore, slightly stronger than the same bonds in the original molecule. The C-C bonds
in {3 position, however, are adversely affected by the described electronic reorgani-
zation. In thermal or mechanical degradation they are found to break preferentially.
Reactions involving the scission of a radicalized carbon backbone chain lead to the
formation of a double bond:
R 1 -CHX-CH 2 -CX-CH 2 -R2 -+ R1-CHX + CH 2=CX-CH 2 -R2.

(3 c<
The considerable amount of energy gained in transforming a -CH 2 -CX- bond

into a double bond (250-280 kJjmol) should decrease the activation energy of such
a reaction correspondingly. Kagiya [14] has derived a number of activation energies,
D, mostly for vinyl polymers. He uses the assumption that D is equal to the heat
of polymerization, Q, and the activation energy Ep of the propagation reaction:
D =Q+ Ep. (4.23)
His values are reported in Table 4.8. Again it must be mentioned that a stress-
induced scission reaction may require an energy of activation which is conSiderably
higher than the listed values of D.
Table 4.8. Activation Energy of Main Chain Scission of Radicalized Chains*
Polymer Activation Energy Polymer Activation Energy

kJ/mol kJ/mol
PE 118 PVF 116

PP 102 PVDF 131
PIB 76 PTFE 147
PS 85 PVAC 118
Pa-MS 95 PMA 98
PVC 87 PMMA 85
PA6 117**
* Ref. 14.
** T. Kagiya, personal communication.
85
III. Form of Binding Potential
In Figure 4.1 it was qualitatively indicated how the potential energy of two bonded
atoms changes with their interatomal distance r. Knowledge of the exact form of
the interatomal potential could - in principle - be obtained from a calculation of the
total electronic energy E of a molecule as a function of r. For polyatomic molecules,
however, E(r) cannot be calculated by any of the mentioned approximation methods
with reasonable accuracy. So the potential functions are described by empirical,
phenomenological functions. These functions will, of course, necessarily utilize
those molecular constants which are otherwise obtainable, e.g. bond length (ro), bond
stretching force constant, and dissociation energy D. One of these functions is the
Morse potential function which has the virtue to lead to a solvable SchrOdinger
Equation:
v = D exp [-2a(r - ro)] - 2D exp [-a(r - ro)]. (4.24)
The Morse constant a can be expressed through force constant and anharmonici-
ty constant x of the oscillator formed by the bond atoms. The Morse function yields
the energy eigenvalues of an anharmonic oscillator which are correct within a size-
able range of amplitudes (in the case of H2 for 0.4 < rlro < 1.6). For large r, how-
ever, V becomes progressively too small [8b].
1. K. S. Pitzer: Quantum Chemistry. New York: Prentice Hall, 1953 a) 153, b) 157, c) 163.
2. R. Daudel, R. Lefebvre, C. Moser: Quantum Chemistry, Methods and Applications.
New York: Interscience, 1959 a) 48, b) 91, c) 3-24.
3. A. G. Shenstone: Phys. Rev. 72,411 (1947).
4. T. Cole, H. O. Pritchard, N. R. Davidson, H. M. McConnell: Mol. Phys. 1, 406 (1958).
5. H. A. Stuart, Ed.: Molekiilstruktur. Berlin - Heidelberg - New York: Springer 1967, 35.
6. R. D. Brown: J. Chern. Soc. 2615 (1953).
7. R. M. Joshi: J. Macromol. Sci.-Chem. A4 (8), 1819 (1970).
8. J. G. Calvert, J. N. Pitts, Jr.: Photochemistry. New York: Wiley, Inc. 1967, a) 815ff, b) 136f.
9. D. W. van Krevelen: Properties of Polymers, 2nd Edition. Amsterdam: Elsevier 1976.
10. A. Yeo Stepan'yan, Yu. G. Papulov, Ye, P. Krasnov, G. A. Kurakov: Vysokomol. Soyed.
A14, 2033-2040 (1972), Polymer Sci. USSR 14,2388-2397 (1972).
11. M. Heskins: Encyclopedia of Polymer Science and Technology, Supplement Vol. 1. H. F.
Mark, N. G. Gaylord, Eds. New York: Interscience 1969, 763.
12. G. Herzberg: Spectra of Diatomic Molecules, 2nd Edit., New York: D. van Nostrand 1953.
13. W. Brenner, W. Kapfer: Encyclopedia of Polymer Science and Technology, Vol. 11, H. F.
Mark, N. G. Gaylord, Eds. New York: Interscience 1969, 763.
14. V. T. Kagiya, K. Takemoto, M. Hagiwara: IUPAC Macromol. Symposium "Long-Term
Properties of Polymers and Polymeric Materials". Stockholm, Sweden, Aug. 30 - Sept. 1,
1976.
86
Chapter 5
Mechanical Excitation and Scission of a Chain
I. Stress-Strain Curve of a Single Chain 87

A. Entropy Elastic Deformation 88
B. Energy Elastic Chain Deformation 93
C. Non-Equilibrium Response 98
II. Axial Mechanical Excitation of Chains 99
A. Secondary or van der Waal's Bonds 99
B. Static Displacements of Chains Against Crystal Lattices 99
C. Thermally Activated Displacements of Chains Against Crystal Lattices 105
D. Chain Displacements Against Randomly Distributed Forces 107
E. Dynamic loading of a chain 108
III. Deexcitation of Chains 111
I. Stress-Strain Curve of a Single Chain
A chain molecule as part of a thermoplastic body is in thermal contact with other

chains and - at room temperature - constantly in motion. The atoms vibrate and
take part in the more or less hindered rotations of groups and even of chain seg-
ments. With no external forces acting all molecular entities try to approach - and
fluctuate around - the most stable positions attainable to them. The action of
external forces causes - or maintains - displacements of the chain from those
positions and evokes retractive forces. Let us consider a chain or a bundle of chains
in thermal contact with the surrounding and at constant volume. The condition
of thermodynamic stability of such a system is that the free energy
F= U- TS (5.1)
assume a minimum. The differential changes of the free energy of the system are
given by the changes of its internal energy U and its entropy S.
dF= dU - TdS. (5.2)
An increase of U can be effected by heat dQ flowing into the system from the
outside or by mechanical work dW done onto the system. The change of entropy
may be split up into a portion -deS which is equal to dQ/T and into the internal
entropy production diS caused by irreversible processes as for instance by stress
relaxation or generation of heat through friction:
87
5. Mechanical Excitation and Scission of a Chain
(S.3)
From Eq. (S.3) one derives the force f acting upon a chain with end-to-end
distancer as:
f=
aw)
(~ V,T =
(aF)
ar V,T +T
(ais)
~ V,T
(S.4)
The uniaxial force on a relaxing chain molecule depends, therefore, on the

changes of internal energy and entropy associated with a change of its end-to-end
distance and on the entropy production term (which is always positive). These three
different force contributions will be examined in the following.
A. Entropy Elastic Deformation
The kinetic theory of rubber elasticity was essentially developed between 1930
and 1943. A detailed description of the development of this theory and its present
state is given by e.g. the classical treatises of Flory [1] or Treloar [2]. The necessary
conditions for the occurrence of rubberlike elastic behavior are the presence of a
long chain molecule possessing internal flexibility (freely rotating links) and the
absence of strong secondary bonds acting between a segment and surrounding seg-
ments of the same or other chains. The stress-strain relation for a single finite chain
then is derived from the distribution of chain conformations. Following the treat-
ment of Treloar it will briefly be recalled which chain properties enter the stress-
strain relations and what levels of force are attainded.
First the conformations of certain idealized chains are considered. For the chain
of n equal links of length Q joined entirely at random (Le. neglecting valence angles
and hindrances to internal rotation) the probability distribution p(x, y, z) to find
a chain end at a point (x, y, z) - if the other end is fixed at the origin - is obtained
as a Gaussian distribution:
(S.S)
The only parameter, b, is the inverse of the most probable chain end separation
and is under the above conditions .J3/nQ2 . The extended length of the chain is, of
course, equal to nQ, whereas the average chain end separation (the "root-mean-
square length"....rrz equals Q yn.
More realistic is the model of the statistically kinked chain with valence angles
and hindrances to internal rotation. Here the mean square length is derived as:
;Z = nQ2 1 + cos ex . 1 + a (S.6)

1- cos ex 1- a
where ex is the semi-angle of the cone of bond rotation and a is the mean value of
cos r/J, r/J being the angk of bond rotation counted with respect to the trans-con-
formation. The ratio r2/nQ 2 is called the characteristic ratio. If the rotation of a
88
1. Stress-Strain Curve of a Single Chain
bond is considered to be independent of the position of other bonds and free with-
in its cone of rotation then a is zero.
A given (hydrocarbon) chain can be treated as a statistically kinked chain with
valence angles and also as freely jointed chain. The root-mean-square lengths and
extended lengths of both models correspond to each other then and only then
when a random link of appropriate length has been used. This "equivalent random
link" of length Qr is rather short and contains only a few chain atoms. For paraffin-
type chains Treloar obtained a length Qr of 0.377 run, for cis-polyisoprene 0.352 nm.
These equivalent random links are about two times smaller than the (shortest) seg-
ment lengths necessary to form a fold of 180 0 •
The entropy S of a single chain is according to Eq. (5.5) and the Boltzmann
relation
(5.7)
where C1 is a constant. An important aspect of the entropy of a chain is - in the

words of Treloar [2a] -: "A chain does not possess an entropy by the mere fact
of having a specified conformation, but only on the virtue of the large number of
conformations accessible to it under specified conditions of restraint". If these
conditions consist only in the location of the end-points of the chain at specified
positions, then the appropriate expression for the entropy is Eq. (5.7). In the
presence of external hindrances to bond rotation the entropy of a chain is not given
by Eq. (5.7) since p(x, y, z) must then be corrected for the number of inaccessible
states.
The retractive force / acting between the chain ends - and in the direction of
the line joining them - is derived from Eqs. (5.2), (5.4) and (5.7). If one considers
a reversible deformation and postulates that all conformational states posses the
same internal energy U so that U does not depend on conformation then
/=2kTb 2 r. (5.8)
The random coil in thermal equilibrium behaves as a spring (of length zero)
with a spring constant 2 kT b2 . The level of force exerted by such a spring with its
ends separated by the most probable distance r = lib is fairly small. For a hydro-
carbon chain of molecular weight 14'000 with an extended length of 125.5 nm
and a most probable chain end distance of 7 nm, this force is 17 . 10- 13 N which
corresponds to a force per cross-section of the chain, //q, of 10 MN/m 2
(l08 dyn/cm2). As one should expect this "chain stress" is very close to the
modulus G =NkT of a rubber network having chains of exactly this length between
crosslink points and for which Gis 17.5 MN/m 2 .
The Gaussian theory considers the number of possible conformations of a chain
having a specified end-to-end distance. A more accurate non-Gaussian statistical
treatment of the random chain is based on the distribution of sin 8 1 , i.e. of the
angle between the direction of a random link and of the end-to-end vector. From
the probability of finding nl links in the range bo8 1 , n2 in bo82 and so on, the
entropy of a single chain is derived [2b] as
89
S. Mechanical Excitation and Scission of a Chain
S = C2 - kn [~13 + Qn _._13_] (5.9)

nQ smhl3
where C2 is a constant and 13 the so-called inverse Langevin function ..2"'-1 (r/Oll.).
Using (5.2) and (5.4) one obtains
f= (~T) ..2"'-1 (r/nQ) (5.10)
which can be expanded into
f =kT
Q)J 3 (2..) + ~ (.:...)
nQ 5 nQ
3 + 297 (r) 5
175 nQ
}
+ . .. . (5.11)
It is of special interest to note that this stress-strain relation can also be derived
from a consideration of the potential energy of the statistical links in a unidirectional
force field. James and Guth [3] and Flory [1] have used the fact that an amount of
work of f . Q(1 - cos ei) must be performed to rotate a link from a position of
e =0 to e =e i ; if the orientational energy -f Q cos e i is distributed according to
Boltzmann statistics then the mean extension of a link in the direction of force is
Zi = Q..2"'(fQ/kT) (5.12)
and the mean total chain end-to-end distance is
r = nZj = nQ..2'" (fQ/kT) (5.13)
which by inversion directly yields Eq. (5.10). It should be pointed out that so far this
result has been derived without introducing - or permitting - any internal or
external hindrances to rotation of a link. The absence of external hindrances also
means that in longitudinal extension or contraction of the chain or of parts thereof
no frictional forces are encountered. The forces exerted by the chain ends are solely
the consequence of the imbalance of exchange of momentum between the chain
and its surrounding.
If a hydrocarbon chain behaves as described by Eq. (5.13) and if such a chain
were to reach a breaking force (of, say, 17 . 10- 10 N) at room temperature then
fQ/kT must exceed 150. For an argument that large the Langevin function is equal
to 1 -kTIfQ andfis:
f= nkT(nQ - r). (5.14)
In the chosen example of a hydrocarbon chain of 125.5 nm extended length,

breaking stresses will only be reached if r is larger than 124.7 nm. In other words only
2 of the 333 random links of 0.377 om length may point in a direction perpendic-
ular to the chain end-to-end vector while all the remaining ones then have to be
fully aligned. Even at this extreme elongation the Langevin function gives a good
approximation of the stress-strain behavior of the randomly kinked chain. This
90
becomes evident by comparison with the so called exact solution of Treloar [2c]
which is purely based on geometrical (and combinatorial) considerations and from
which one derives for extreme elongations:
f= -kT dl2np/dr =-kT/r + (n-2) kT/(nQ-r). (5.15)
The force-elongation relations (5.8), (5.10), and (5.15) have been derived under
the condition that all conformational states have the same internal energy, that all
states are accessible, and that transitions between different states are easily accom-
plished. No chain molecule - and especially not within a solid - does fully comply
with the above conditions. For one thing there are the intramolecular interactions
(between nearest and second nearest atoms of one chain) which cause the potential
energy of the chain to be a function of the angle of rotation with distinct relative
minima in more than one position. All this is well known and documented (e.g. 1-5)
and will not be discussed here. The direct or indirect consequences with respect to
the retractive forces between (PE or PA) chain ends are the following:
- the internal energy of (in-plane) trans-conformations is lower than that of (out-
of plane) gauche-conformations
- the probability to find a conformation with internal energy LlU is weighted by
the Boltzmann factor exp (-LlU/kT)
- the average end-to-end distance increases with a according to Eq. (5.6), the
length Qofthe equivalent random link also grows in proportion to (1 +a)/(1 -a),
while the number n of random links in a chain of finite length decreases
the force between chain ends at their most probable distance (r = lib) is in-
versely proportional to r and thus also to the square root of the characteristic
ratio, the force is the smaller the larger the characteristic ratio is.
The characteristic ratio of PE at room temperature is - according to Abe
et al. [4] - 8.0 as compared to a value of 2 calculated under neglect of hindrances
to bond rotation. The force between PE chain ends, therefore, is by a factor of 2
smaller if the effect of the potential energy on chain conformation is taken into
consideration. For polyamide chains the detailed calculations of Flory and Williams
[5] led to characteristic ratios of 6.08 (PA 6) and 6.10 (pA 66) in excellent agree-
ment with the experimental value of 5.95 for PA 66 found by Saunders [6]. The
calculated force between PA chain ends, then, will be smaller by a factor of about
1.7.
A second cause to invalidate Eq. (5.7) are strong intermolecular interactions
between chain segments which do have the effect that only a restricted number of
conformational states are accessible to a chain segment, that transitions between
different states may not be independent from each other, and that energy is required
to activate these (cooperative) transitions. In recent years considerable progress has
been made to apply the methods of statistical thermodynamics also to systems
subject to the above restrictions and to explain e.g. their dynamic stress-strain
behavior [7] or their melting and crystallisation [8]. With respect to chain loading
one is interested in the entropy changes associated with the elongation of a highly
extended chain. In the following LlS will be calculated for PE and PA chain segments
of 3 to 6 nm length.
91
It is convenient to start with Pechhold's [7] treatment of the polyethylene

molecule. He derived the partition function of the rotational isomers of a chain
with n independent C-C bonds. The chain is allowed to rotate around its C-C
bonds, it will, therefore, encounter three minima of potential energy of which the
absolute minimum belongs to the extended trans-conformation (t). Two relative
minima at an angle ± 1200 out of plane allow two gauche-conformations (g,g)
with a conformational energy of + ~U. The kink-isomers are the subclass of the
rotational isomers where only n/2 of the bonds - and no adjacent ones - may have
a g or g position. For kink isomers entropy, internal energy, and free energy as a
x
function of the average concentration of gauche-conformations are derived [7]
as:
x = exp (-~U/kT) / [1 + 2 exp (-~U/kT)] (5.16)

S = - 0.5 nk Qn (1 - 2x) + oX ~U/T (5.17)
U =nx ~U (5.18)
F = 0.5 nkT Qn (1 - 2x). (5.19)
The group of kink isomers still comprises chains with quite different deviations
from a planar zig-zag conformation, the simplest of those kinks being ttgtgttt
(2gl kink), tgttgt (2g2 kink), tgtgtgt (3g2 kink). Each kink causes a contraction
in longitudinal direction which is a multiple & of Y2f3a with a the length of a C-C
bond (pechhold accordingly denotes a kink containing q gauche positions and
causing a contraction & as a qg ~z-kink). A chain with n C-C bonds and mi qgi-
kinks has a projected length in z-direction of (n - imi) Y2f3a. Vice versa a pro-
jected length corresponding to that of n C-C bonds can be realized as well by an
extended chain with n C-C bonds, by a chain with n + 1 bonds containing a 2gl
kink, by chains with n + 2 bonds containing either two 2g1 kinks or a 2g2 kink or
by other appropriate combinations. The statistical weight of each conformation is
given by the number of ways that conformation can be realized and by the Boltz-
mann factor of its conformational energy. The by far leading term in the partition
function for a given r is always that listing the number of possible conformations
of chains containing only 2g1 kinks. The number of ways of arranging ng/2 pairs of
g and g bonds on n/2 available positions clearly is
n g /2
Z (n, ng) = 2ng/2.1T
1=1
(n-2 - 2i + 1) exp (-2i~U/kT) . (5.20)
In Table 5.1 the statistical weights ofPE chains with a gauche-concentration of

up to 20% are given according to Eq. (5.20). Also listed are the statistical weights
of 6 PA chains subject to the condition that the two gauche positions of anyone
gtgt kink are confined to the (six) C-C bonds of one monomeric unit and that the
carbonamide group adheres to a trans-planar conformation [5]. As potential energy
~U of a gauche position the "preferred value of 0.5 kcal/mol" [5] has been used.
The change of Qn Z (n, i) with r is essentially the same as the change of the
complete partition function (which accounts for mgi-kinks of all types). One can cal-
92
Table 5.1. Statistical Weight of (n, ng) Conformations

Polyethylene chains, l> U = 0.5 kcal/mol
n ng = 0 2 4 6 8 10
38 6.80 41.1 217 985 3722

40 1 7.18 46.1 261 1282 5329
42 1 7.55 51.4 311 1642 7444
44 1 7.93 56.9 366 2072 10174
46 1 8.31 62.8 427 2580 13646
48 1 8.69 68.9 495 3177 18000
6 Polyamide chains, l>U of gauche positions of C-C bonds 0.5 kcal/mol
n ng = 0 2 4 6 8 10
35 1 5.67 25.7 87.3 198 224

42 6.80 38.5 174.6 594 1346
49 7.93 53.9 305.6 1385 4710
56 9.07 71.9 489.0 2771 12560
culate, therefore, from the data of Table 5.1 the entropy elastic contribution to the
retractive force as - kT Ll Qn (Z)/Llr. The energy elastic contribution, fe, derives
here from the change of conformational energy and is LlngLlU/Llr. For the chosen
examples of a PE chain with 40 C-C bonds, a 6 PA chain of 6 monomer units and
gauche energies LlU of 0.5 and 0.75 keal/mol one obtains the force contributions
listed in Table 5.2. For comparison the forces derived for the statistieally kinked
chain according to the Langevin and the "exact" approximation are also given. It
is immediately apparent that in kink isomers the energy-elastic contribution fe is
of the same order of magnitude - and opposite sign - as the entropy-elastic force.
The molecular stresses, 'l!, are obtained by dividing the axial forces f through the
molecular cross-section q, whereas the elastic moduli are understood as:
E = r Llf/ Mq. (5.21)
This investigation may be summarized by saying that the retractive forces ex-
perienced by highly extended chain segments at room temperature due to confor-
mational changes are small in comparison to the ultimate forces that can be carried
by a chain. External hindrances to segment rotation have not been treated here. At
present is suffices to state that they will only further diminish the -already - small
entropy elastic forces.
B. Energy Elastic Chain Deformation
So far the energy-elastic deformation of the (assumedly rigid) chain skeleton has
not been considered. In response to axial forces the chain skeleton will deform,
93
\0 :"
~
s:::
(Il
n
::r
'2."
n
e.
tTl
~
a·
....
'o·"
Table 5.2. Entropy-Elastic Forces and Moduli ::l
::l
'"
0-
Approach Eq. L r fe f E Vl
n
lO-lON lO-lON MN/rn2 ~.
A A '" GN/rn 2
5'::l
Langevin (Q = 0.377 nrn) (5.14) 49 45.2 0 1.43 843 5.5 ...,0
Exact (Q = 0.377 nrn) (5.15) 49 45.2 0 1.20 712 4.6 (")
'"
PE-Kinkisorner (~U= 0.50) (5.20) 50.3 49 -0.55 0.10 55 1 ::r
PE-Kinkisorner (~U = 0.75) (5.20) 50.3 49 -0.83 -0.45 -247 (O.9) S·
'"
PA-Kinkisorner (~U = 0.50) (5.20) 52.1 50.8 -0.55 0.08 49 1.9
PA-Kinkisorner (~U = 0.75) (5.20) 52.1 50.8 -0.83 -0.47 -278 (1.6)
PE bond stretching and bending: 23.8 14000 (235)
Table 5.3. Stiffnesses of Molecular Deformation Modes
Mode of Deformation Polymer Calculated Derived from Ref.

Modulus IGN/m 2 ]
Hindered rotation (helix) Polyesters 4-7 X-ray diffraction on various polyesters 12

Hindered rotation of kinked chain PE 2.6 n/ng Eq. 5.21, sinusoidal rotation potential
Pure bending of C- C bonds PE 80 12
Pure stretching of C-C bonds PE 740 12
Stretching of zig-zag chain PE 182-340 Force constants 9
PE 182,299 Urey Bradley force field 13
PE 297 Minimum energy electron wave function 11
PE 272 Deformation of rigid valence tetrapods 47
PA 6 180-200 Spacing of (070) planes 46
t-'
PA 6 157-196 Force constants 9 Vl
....
...
PA6 244 Urey Bradley force field 13 ~
PA 66 181 Urey Bradley force field 13 ~
~
Crystal lattice deformation PE 235 Spacing of (002) planes 9
S·
'"
PA 6 (0< form) 20,24.5 Spacing of (020) planes 46,9 n
:::
...<
0
'"...,
Vl
'"
S·
~
'"
n
::>"
\0
VI S·
'"
however, through hindered rotation, bending, and stretching of main chain bonds
(9). In Table 5.3 the stiffnesses characterizing these individual modes of deforma-
tion and their combination in a loaded chain are listed. The C-C bond bending and
stretching moduli have been calculated from the corresponding force constants. The
modulus for the hindered rotation of a kinked chain is derived from the rotational
potential of CH 2 groups and the associated axial chain elongation as formulated
by Pechhold [7] who used an intramolecular rotational barrier of 2 kcal/mol above
gauche energy. The resulting modulus is a function of the angle of rotation and
ranges between 4 and 7 GN/m2 in good agreement with a value of Bowden [10]
of 5.7 GN/m2. In axial stretching the hindered rotation of gauche bonds contributes
to axial elongation, the chain stiffness thus decreases with the concentration of gauche
bonds.
Longitudinal chain moduli, defined as in Eq. (5.21), have been determined
theoretically and experimentally in a number of ways:
through calculation from spectroscopically determined bond bending and stretch-
ing constants
from the Raman shifts of extended chain hydrocarbons
from x-ray observations of load-induced changes of the unit cell dimensions in
highly oriented polymers.
These methods, their results, and the arguments to explain a certain disagreement
between the numerical results have been reviewed by e.g. Sauer and Woodward [9].
More recently the direct calculation of the total ( electronic) energy (of a PE chain)
with changing atomic geometry has been attempted by Boudreaux [11]. In his self-
consistent field molecular orbital (SCF - MO) calculations he searched for "that
set of numbers which, when used as coefficients of the assigned atomic orbitals,
produces an approximation to the wave function which minimizes the total energy".
Hahn et al. [14] presently investigate a model based on the elastic interaction of
essentially rigid sp3 valence tetrapods (shell-model).
From Table 5.3 it is apparent that the weakest mode of skeleton deformation
(4.7 GN/m 2) is that derived from the hindered rotation of (non-extended) bonds
whereas a fully extended chain segment has a 50 times larger stiffness. The elastic
modulus of a partially extended kinked chain, then, is:
I/E = ~ xdEi (5.22)

i
where the Ei are the stiffnesses of the individual (trans-planar, kinked, or helical)
segments of a chain and the Xi are the corresponding (volume) fractions.
One is now in a position to determine the change of the free energy F of a
partly extended chain as a function of the end-to-end distance r. As an example
the quasi-static deformation of PE segments will be treated within the frame of
the bond-bending and stretching model. The minimum free energy of a segment
containing n C-C bonds and nk 2g1 kinks is attained at an end-to-end distance
r =(n - nk)V2f3 a. This minimum free energy is equal to nk ~U-RT Qn Z. From
the statistical weight of the (n, nk) conformations of PE segments with n = 40
(Table 5.1) the minimum free energy values have been calculated. They are indicated
in Figure 5.1 at the proper chain-end distances. If the chain ends are axially dis-
96
F
kJ/mol
o
\\ I 3 2
\
I
I I
\ I /
\ \ I /
\ \ I I
,., \ \
\ \ I
I I
I
....
Cl
~
co '-.l / I
.
!t
I
I
I \
r nm
_ 1 L __ _ _ _- L__- L__ ~~~~L_~ __ ~ _ __ L_ _ ~
4.4 4.S 4.6 4.7 4.8 4.9 S.O nm

Distance of chain ends
Fig. 5.1. Free energy of uniaxially deformed polyethylene chain segment. Number of C-C-
bonds n = 40, number of kinks nk = 2 ng, extended length L = 5,03 nm.
placed from these equilibrium positions energy elastic forces arise which have an
unsymmetric potential. In stretching completely extended sections the trans bonds
are deformed in the plane of the zig-zag chain according to the chain modulus E str •
The gauche bonds are performing a hindered rotation out of the zig-zag plane
(Erat). The stretching modulus E of the kinked segment then is derived from
Eq. (5.22). The chain is the stiffer the fewer gauche bonds it contains. Employing
the rotation potential mentioned above [7] and an extended chain modulus of
200 GN/m 2 the tensile sections of the free energy curves have been calculated. The
presence of only 5 kinks softens the chain segment quite noticeably (E decreases
to 9.5 GN/m 2).
The axial compression of an extended chain differs in so far as the trans bonds
now aquire the freedom to perform a hindered rotation out of the zig-zag plane.
This freedom is not exercised in tension because it would be energetically less
favorable than bond bending and stretching. The axial compression modulus of a
zig-zag chain is essentially determined by the hindered rotation mode of deforma-
tion, it corresponds to the extension modules of a highly kinked chain (nk = n/4)
and has been taken to be 5 GN/m2.
For each conformational state an asymmetric parabola is obtained. In quasi-
static straining an individual, kinked chain will tend to assume that state of minimum
free energy accessible to it. Transitions between different conformational states will
occur (solid line in Fig. 5.1). The transition to a more extended conformation will
cause an immediate release of axial stress for the chain segment in question - and
97
thus constitute a thermodynamically irreversible process. Macroscopically the step-

wise and uncorrelated release of stress in a large number of chains is felt as stress
relaxation. The random thermal motion of chain atoms and the existence of ex-
ternal hindrances to bond rotation are the reason that the deformation potentials
of individual chain segments deviate from each other and from the solid line - as
arbitrarily indicated by the heavy dashed line. The deformation potential shown
in Figure 5.1 essentially represents also that of other (planar) zig-zag chains such
as PA 6 or 66 which have the same kind and concentration of gauche bonds and
about the same length.
C. Non-Equilibrium Response
The stress-induced conformational transitions of a kinked chain as represented by

the solid line in Figure 5.1 are thermodynamically irreversible, internal energy is
converted into heat. At this point one is interested in the time-constant of the tran-
sition process: if it is small compared to the time within which chain stretching is
accomplished then the stress-strain curve will not deviate very much from the solid
line, if the time constant is extremely large then transitions may be considered to
be impossible, the chains deform energy-elastically. In the intermediate region of
time constants, however, the largest stresses of not fully extended chains will depend
on the rate at which stress releasing conformational transitions take place. A detailed dis-
cussion of this effect would require a consideration of the geometry and interaction
of molecular chains, of the principles of irreversible thermodynamics [15], and analy-
sis of the potential of the secondary or van der Waal's bonds between segments
[16]. It would then lead to a description of the anelastic deformation of polymers
which is not intended in this book. Nevertheless one is still interested in some in-
formation with respect to the rate of stress relieving transitions between different
kink-isomers. Since any transitions leading to the motion of just a single kink gene-
rally do not cause any axial chain elongation one has at least to consider the simul-
taneous activation and annihilation of two kinks. Such process involves the rotation
of four gauche bonds and the rotational translation of the segment between the
kinks, the energy of activation can be estimated to be 8 kcal/mol for the rotation
of gauche bonds [7] and 0.5 to 1.2 kcal per mol of rotating CH 2 groups to overcome
the intermolecular barriers [17, 18]. A tgtgttg tgt-kink requires, therefore, an acti-
vation energy of 11 to 15.2 kcal/mol to transform into the all-trans conformation.
One may assume that in stress-free PE chains such transformations do occur with
a rate of at least 1 S-l at a temperature somewhat below the melting point, i.e. at
400 K. One may then calculate the rate of this process at 300 K from Eq. (3.22)
as 0.0018 s-l . If the chain is strained the activation energy for segment rotation
only decreases and the rate of stress relieving transitions increases [19]. With such
a time-scale one may consider stress-induced rotational transitions as being easily
accomplished before a chain segment reaches elastic energies sufficient for chain
scission at room temperature, i.e. more than 25 kcal/mol.
It remairls the entropy production term in Eq. (5.4) to be discussed. Under
the conditions of this section - quasi-static axial loading of the ends of a segment
and determination of the ensuing retractive forces - this term is essentially zero.
98
II. Axial Mechanical Excitation of Chains
A. Secondary or van der Waal's Bonds
In Section A the stress-strain behavior of a single chain segment loaded pointwise at

its ends has been discussed. In (uncross-linked) thermoplastic polymers, however,
large axial forces cannot be applied pointwise to a main chain bond but only be
built up gradually along the chain through - the much weaker - intermolecular
forces. The forces between molecules are a superposition of the close range (nucle-
onic) repulsion and the (electronic) van der Waal's attraction, which includes elec-
trostatic forces between ions, dipoles or quadrupoles, induction forces due to polari-
zation of atoms and molecules, and the generally most important quantum-chemical
dispersion forces. The van der Waal's attraction is the cause of the solidification and
crystallization of high polymers, it is theoretically quite well understood and has
been treated in detail by Langbein [16]. With reference to this work and the general
references in Chapter 1 it can be stated that the secondary forces are not saturated
and are nondirectional, i.e. not confined to precise locations of the neighbor atoms
at e.g. tetrahedral bond angles. To the extent that these assumptions are valid the
potential of the intermolecular forces acting on a chain or segment can be replaced
by the sum of the interaction potentials of all appropriate pairs of atoms. The pair
potentials contain an attractive term - which is theoretically derivable and decreases
with the 6th power of the interatomal distance [16] and a repulsive term for which
only semi-empirical expressions can be given. The total energy of intermolecular
interaction, i.e. the cohesive energy of a solid then is to be understood as the sum of
the pair interactions. In thermodynamic equilibrium a chain segment attains' a position
such that the free energy of interaction is a minimum. If a chain segment is displaced
with respect to this position the free energy is raised and retractive forces ari&e. Such
axial displacements are realized in quasi-static pulling a chain segment which is part
of a crystal lamella against the crystal lattice or of extended chains against a glassy
matrix. Dynamic chain stretching occurs in the shear deformation of molecules in
solution or the melt. In the following it will be discussed which level of axial chain
stress can be attained in these cases, that is in the loading of primary bonds through
secondary or van der Waal's forces.
B. Static Displacements of Chains Against Crystal Lattices
The first investigations of the loading of chains displaced with respect to their
equilibrium positions within a crystal lattice have been carried out by Chevychelov
[20]. He used a continuum approach, i.e. replaced the string of discrete chain atoms
by a continuous mass distribution. Each mass point experiences a periodic repulsive
crystal potential if displaced from its equilibrium position. Later Kausch and Lang-
bein [21] and Kausch and Becht [22] extended these calculations to treat the static
and dynamic interaction of chains of discrete atoms with arbitrary periodic poten-
tials.
99
The model used in these investigations is represented by Figure 5.2. It consists

of a tie molecule (t) which forms part of a crystal lamella ( c), leaves it perpendicular.
ly to the crystal fold surface, (I), extends straightly through an amorphous region
(a) and enters the adjacent crystallite of which it forms a part. The crystal boundaries
are assumed to be ideally sharp. No interaction between the tie chain and the amor·
phous material outside of the crystallite is taken into account.
, . .,:'!-----+'!,)
, ,
: ! Fig. 5.2. Model of tie molecules (t) con·
, c
I necting folded chain crystal lamellae (c).
,
I
a = amorphous region, Lo = length of
,
, amorphous region, L = contour length of
tie molecule, I = crystal boundary.
A tie molecule thus embedded in two different crystalline layers is streched if

the crystal lamellae are separated by macroscopic stresses. Because of the large elastic
modulus of the chain molecule very high axial stresses are exerted onto the chain
if it experiences the same strain as its amorphous vicinity. This high axial stress
tends to pull the tie molecule against the intracrystalline potential and out of the
crystal lamella.
In the following calculations the assumedly rigid solid is represented by an
interaction potential the depth of which corresponds to two times the average
(lateral) cohesive energy U with which the chain segments are bound within lamellar
crystals or amorphous regions. Upon axial displacement in z·direction ( chain axis)
a chain will rotate and translate along a temperature· dependent minimum energy
path. For polyethylene (PE) Lindenmeyer [17] has given the maximum potential
energy to be overcome by a CH 2 group in twisted translation as LlUtt = 500 cal/mol
at room temperature. The potential energy of a CH 2 group within the amorphous
phase can be determined from the difference of the enthalpies of crystalline and
amorphous PE at absolute zero temperature, Ham, 0 - Her, 0 = 661 cal/mol [23].
It should also be taken into account that the internal energy of a partly oriented
PE may be smaller than that of an amorphous one. The energy difference will be
comparable to the enthalpy difference (Ham - Hder), which was estimated by
Fischer and Hinrichsen [24] to be 140 cal/mol. As function of its z'position within
an otherwise undisturbed crystal a CH 2 group would encounter a potential energy
as shown in Figure 5.3.
Each point along a tie chain is characterized by its distance z from the crystal
boundary, where z refers to the unstressed situation. The displacement of the molec'
ular point z under stress is denoted by u = u (z) and the crystal potential per unit
100
amorphous
z
Q.
"en2
N
:I:
u
I
J'~~
2U tOh
"
"0
,., -1
..
!!'
~ undeformed
_._._0-
:g 2(H am; H,te,)
i: deformed
20 dU"
Q.
-2
Fig. 5.3. Lattice potential in displacement of a CH 2 group in crystalline and amorphous poly-
ethylene (after [221). Vcoh = cohesive energy per CH 2 in PE crystal, AV tt = potential maximum
in twisted translation, Ham, 0 = enthalpy of PE crystal at 0 K, Ham - Hdef = difference of
enthalpies of undeformed and deformed amorphous phase of PE.
length by V(u). The axial stress, s, acting along the chain decreases with increasing
Izl due to the forces exerted on the chain through the crystal potential V(u).
Equilibrium of forces at each section of the chain is established if
~ = dV(u) (5.23)
dz du
This relation implies no assumptions with respect to the form of V(u) or the
response s(u) of the chain.
One does assume, on the other hand, the tie-chain segments within the crys-
tallites to resemble an elastic spring, i.e. one equates the axial stress s to the local
chain deformation
s = Kdu/dz (5.24)
yielding
d2 u _ dV(u)
K -- ---- (5.25)
dz2 du
Here K is the constant of chain elasticity (not normalized with respect to the
chain cross-section).
For a semi-infinite crystal and a sinusoidal potential V(u) = v(1 - cos K u(z))
Chevychelov integrated Eq. (5.25) to obtain
2"1 (dU)2
dz =A - V
K cos K u (5.26)
101
and, after having determined A = v/" from du/dz = 0 for zero displacement,
: =2 .JV7i< sin K u/2 (5.27)
and
4
u(z) = K arctan {C exp [(K v'V/K) z]}. (5.28)
The integration constant C is tan Ku(0)/4 and K is 21T divided by the intra-
chain repeat unit length d. For a polyethylene chain with a spring constant" of
34 nN Eqs. (5.28) and (5.24) have been evaluated [22]. The crystal potential depth
v has been derived as 0.0138 oN from the cohesive energy density 02 through
(5.29)
where NL is Avogadros number, M the molecular weight of a repeat unit, and p

the density. In static mechanical equilibrium the stresses and displacements of tie
chain atoms within the crystallite are a unique function of the axial tension sa exist-
ing at the crystal boundary (z = 0). In Figure 5.4 tension and displacement of aPE
chain within a PE crystallite are shown for maximum axial chain tension So equal
to 1.372 oN. For any force larger than So no static equilibrium can be established
within a defect-free crystallite. It should be noted that this force which pulls out of
the crystal a chain of full crystal length is 15.8 times larger than the force" v d
which is necessary to pullout a single monomer unit. If the tension So is divided
_s_
~--~I
nm
nN
/
/
. I
I :
o~~ ________ ~ ________ ~ ________ ~ ________ ~o
-5.04 -3.78 -2.52 nm -126 z o

distance from crystal boundary
Fig. 5.4. Decrease of tension and displacement of a highly stressed tie chain within a polyethylene
folded chain crystal lamella (after (221).
102
by the chain cross-section (0.1824 run2) one arrives at the maximum axial stress
which a perfect PE crystallite is able to exert on a tie molecule: 7.5 GN/m2. As can
be seen from Figure 5.4, the maximum mechanical excitation of a chain penetrates
about 5 run into the crystallite. At a distance of 6 run from the crystal boundary it
is smaller than the average thermal amplitude at room temperature, which is about
0.008nm.
Kausch and Langbein [21] extended the above calculations to various boundary
conditions at the lower stress end, -Z, of the chain. If the chain terminates at -Z
or continues into the adjacent amorphous region, it experiences no or only very weak
axial forces. This case is covered by putting A < v/". If, on the other hand, a stressed
tie-chain folds back at -Z, any axial displacement there implies a deformation of
the fold. Such a deformation entails a considerable distortion of the crystal lattice
and requires strong axial forces. In this case the integration constant A generally
becomes larger than v/". After having derived the equations necessary for describing
the elastic interaction between a chain and a crystal - represented by one unique
potential- for arbitrary boundary conditions the authors [21] took into account
also the fact that in a crystal containing hydrogen bonds - like that of 6-polyamide -
the intermolecular interaction differs in different crystal regions. It is much stronger
at the sites of C=O and N-H chain groups than it is at the sites of CH 2 groups. The
problem of calculating the elastic displacement of a continuous chain passing through
alternating regions of strong and weak attraction can be solved through alternate
application of appropriate boundary conditions. In this case the boundary condition
is that chain displacement and chain tension are continuous at all points Zj where
the potential changes from weak attraction (v = vd to strong attraction (v = V2)
or vice versa (refer to Fig. 5.5 for notation). Here it was assumed that the difference
%3 %2 %,
I I I I
I I I I
I AC~
z x I A, • v,/x I I
I I I I
I Vz .K z I v, .K, I vz.K z I
I I I I
I I I I %
I I I I •
-1.736 -'.498 -0.868 -0.630 nm
I I 1
I CO - NH I - CHz -CH z - CHz -CHz- CH z - I CO - NH
I I I
I strong I weak interaction I strong
I I I
Fig. 5.5. Geometry and notations used in calculating stresses and displacements of 6 polyamide
chains.
between the cohesive energies Ucoh ofPA (18.4 kcal/mol) and PE (6.3 kcal/mol) is
caused by the presence of hydrogen bonds between the carbonamide groups of
neighboring molecules within the grid planes of the ( monoclinic) P A crystallites.
This excess energy contributes to the intermolecular potential V2 acting between
carbonamide groups. The potential V2 thus is (12.1 + 0.50) kcal/mol, which is
equivalent to 0.369 nN.
103
Chain tension and displacement of a 6-PA tie molecule within a 6-PA crystal
are shown in Figure 5.6. The strong attractive effect of the hydrogen bonds results
in a rapid drop of chain tension at the position of the carbonamide groups. The de-
crease of both tension and displacement is much more rapid than in the case of PE.
At a distance of 2.1 nm from the crystal boundary the displacement has already
dropped to the average level of thermal vibration at room temperature. With the
crystal boundary positioned - as shown in Figure 5.6 - at the end of a (-CH2 -)5
segment the maximum chain tension is So = 3.94 nN. This tension is only 1.7 times
the force necessary to remove one carbonamide group from the crystal, i.e. 59%
of the maximum chain tension is expended to break the hydrogen bonds of the
,---- 0.20
nm
s
nN
0.15
I
/
0.10 ~
1
0.
I
I c
~
II I
/ ~
I
/
0.05
¢=Ji Q
/
C
/
/
. /
o H-- 1" Fig. 5.6. Decrease of tension

and displacement of highly
stressed PA 6 tie molecules
CO NH CO NH CO NH CO NH
- 3.472 -2.604 -1.736 nm -0.868 Z 0 within folded chain crystal
distance from crystal boundary lamellae (after [22]).
first CONH group. It should be noted that the value of So depends upon the geometric
arrangement of the atoms at the crystal boundary. With a carbonamide group at the
crystal boundary the maximum tension So is about 20% smaller than in the case
where the crystal boundary runs through the middle of the (-CH2 - h segment [21].
If So is divided by the molecular cross section (0.176 nm 2) a stress of 22.4 GN/m 2
is obtained, which may be termed the static strength of a perfect (monoclinic)
6-PA crystal.
The above strength values have been derived for the interaction between a periodic
potential and a continuous elastic chain. Kausch and Langbein [21] have shown, how-
ever, that there are only insignificant changes if the model using a continuous chain
is replaced by the more accurate one using a string of discrete atoms.
104
C. Thermally Activated Displacements of Chains· Against Crystal Lattices
As stated repeatedly crystal and chain are in thermal motion; the chain atoms, there-
fore, vibrate around the new equilibrium positions z + u given e.g. by Eq. (5.28).
The effect of the correlated or uncorrelated motion of the chain atoms on the inter-
molecular potential energy can be calculated if the dispersion of vibrational ampli-
tudes along the chain is known [22]. A chain subjected to maximum tension So
is not resistant to translation. Even the smallest thermal vibration will cause further
displacement of the chain and a decrease in axial tension at the boundary. The point
of maximum tension will rapidly propagate into the crystallite (Fig. 5.7). The "dis-
CHz - CH z
Sa
nm 0.5 Z
distance from crystal boundary
Fig. 5.7. Schematic representation of a stress-induced defect. The point of maximum chain
stress travels into the crystal as the leading chain group is pulled out of the crystal (after [22]).
location" is fully developed once Ua has reached the value of d. The energy Wd of
the defect introduced into the crystal by displacement of the atoms of one chain
is easily calculated as the sum of the potential energies of the chain atoms and of
the elastic chain energy. Both are a function of external tension. In Figure 5.8 the
dislocation energy has been plotted as a function of displacement U a and tension
sa at the crystal boundary. The energy of a completed defect (displacement of the
chain by d) in PE amounts to 32.85 kcal/mol. At tensions smaller than So the dis-
placements Ua are smaller than d/2, the defect energy is smaller than Wd(SO)' and
the defect has a stable position. At displacements larger than d/2 the system becomes
unstable since the chain tension decreases with increasing displacement ua .
As shown in Figure 5.8, transitions between the stable and the unstable state
can be caused by thermal activation. The required activation energy Ud is given by
(5.30)
In Table 5.4 the activation energies required for translation of CH 2 -CH 2 seg-
ments are listed for various displacements U a and tensions sa. One observes that
at chain tensions which are as little as 4% less than the maximum tension still an
activation energy for chain translation of 9.8 kcal/mol is required. One may safely
105
tension
s• 1nN o
.!!£!!!. nm
mol
30
'c
o 20
~ a
~ C
"
'6
'0
>-
activation energy Ud
.
0.126 E
g
e>
'"
c ~
'6
'"
10 -
o ~
o 0.1 0.2 Ua 0.3 nm
tie chain displacemen1 at crystal boundary
Fig. 5.8. Energy Wd of a chain displacement defect in a polyethylene crystal as function of tie-
chain displacement u a at crystal boundary (after [221).
Table 5.4. Activation Energies for Chain Translation in PE
Sa
10- 10 N 10- 20 J kcal/mol
o o 23.08 33.0
0.104 1.79 22.89 32.7
0.195 3.38 22.38 32.0
0.303 5.18 21.40 30.6
0.412 6.91 20.00 28.6
0.494 8.12 18.69 26.6
0.589 9.41 16.90 24.2
0.699 10.71 14.60 20.8
0.824 11.93 11.68 16.6
0.914 12.63 9.39 13.40
1.009 13.18 6.88 9.82
1.108 13.56 4.19 5.98
1.158 13.67 2.80 4.12
1.209 13.72 1.40 2.06
1.260 13.72 o o
106
state, therefore, that even under consideration of longitudinal chain vibrations

chain translation will occur at chain tensions close to the specified values.
D. Chain Displacements Against Randomly Distributed Forces
The rupture of chains in amorphous matrices or in solution clearly indicates that

large axial forces can also be reached in the absence of periodic potentials. Again
forces derive from the displacement of chains or of parts thereof with respect to
the surrounding matrix. If one divides a non-extended chain into small sections of
identical projected lengths one will obtain sections of quite different conformation.
As well the sectional chain moduli Ei as - most probably - the interaction poten-
tials Vi = -Wi' Uj2/2 will vary from section to section. To obtain the axial displace-
ment of an arbitrary sequence of chain sections an interpolation given by Kausch
and Langbein [21] may be used who have found that the changes of chain stress a
and displacement u within a heterogeneous matrix are interpolated by an exponen-
tiallaw:
u = Uo exp "1 z/L (5.31)
(5.32)
The chain stress is derived from Eq. (5.24) replacing the chain elasticity constant
I<. by the elastic constant of a series of elastic elements
(5.33)
The above equation shows quantitatively that large chain stresses will only be
obtained ifthe interaction force constants Wi are of about the same order they are
in a crystal and if the elastic moduli of the chain sections are large throughout. One
"weak" section will greatly enhance average displacement and reduce the stress. It
should also be mentioned that it is not the chain length L and the absolute number
n of interaction sites which raise the chain stress but the intensity of the interaction
forces per unit chain length.
As an example of the static excitation of a chain under these conditions a nearly
extended chain of end-to-end distance L embedded in a glassy matrix may be con-
sidered. If the matrix (index m) and the chain (no index) were homogeneously
strained in chain axis direction they would experience the following stresses:
(5.34)
and
(5.35)
107
Due to the different elastic response the free chain would have to act against
the lattice potential until chain stress and lattice interaction were in equilibrium at
any point of the chain. This interaction can be described by identically the same
mathematical formalism developed above. Since the chain ends are assumed to be
free they cannot bear any axial stresses. This condition will exactly be met if one
superimposes the hypothetical stress a (Eq. 5.35) by a compressive stress according
to Eq. (5.33) where
Uo = €mL/r. (5.36)
The largest stresses will be found in the center section of the strained chain and
here the superposition yields:
€m L
a(L/2) = ~ L-!E. [1 - exp (-'Y/ 2)]. (5.37)
1 1
The distribution of axial stresses along the chain is characterized by the two
stress-free chain ends, two terminal sections of length L/r where the stress increases
with a rate determined by 'Y/L, and a central chain segment subjected to the maximum
stress given by Eq. (5.37). The continuous straining of a chain segment through in-
teraction with (harmonic) lattice potentials - as required for the terminal sections -
is limited to a section of finite length because the largest displacement u may not
exceed the range of validity of the intermolecular potential which is of the order
of 0.1 nm. For breaking stresses of 20 GN/m2 to be reached under such a condition
the constant 'Y/L must be larger than 0.01 nm and the average chain modulus at
these loads must be that of a fully extended chain; since the displacements of PE
and 6 PA chains within their rigid crystal lamellae led to 'Y /L-values of 0.004 and
0.015 nm respectively [22] one is forced to the conclusion that the displacements
of extended chains within the less tightly packed random aggregates of misaligned
chains can only lead to smaller than the above values of 'Y/L. If chain scission were
to occur under the above static conditions it would require a sufficiently stable
physical crosslinking of the (then non-extended) terminal chain sections to the
matrix through folds or kinks.
E. Dynamic loading of a chain
So far the quasi-static interaction between a chain and a surrounding matrix has
been investigated. It has been indicated that the level of axial forces to be obtained
in that manner is limited by the occurrence of chain slip. In dynamic loading these
force levels can be exceeded if frictional forces or forces of inertia become effective.
If the chain displacement does not occur in static equilibrium and not through
a single thermally excited step then the translation will not occur reversibly along
a path of least energy but will require more energy than in the previous cases. The
rate-dependent energy expended per unit distance of forced translation of a chain
segment i is equivalent to a frictional shearing force ri. The response of chains to
108
II. Axial MechamCai t.xcltauon 01 \.-mllns
shearing forces in solution has extensively been investigated and discussed [25]. A
number of different molecular theories of the viscoelastic behavior of polymer chains
in solution have been developed. These predict a relation between e.g. molecular
weight M (or degree of polymerisation P), solvent viscosity 1]s, intrinsic viscosity
[1]] = lim (1] -1]s)/C1]s, molecular friction coefficient ~o, and mean square end-to-end
distance r5. Following the bead spring model of Rouse [25] one obtains
(5.38)
°
The quantity ~ represents the frictional force encountered by each monomeric
unit within the chain which is moved with unit velocity through its surrounding.
The sectional force fi' therefore, is:
(5.39)
In the case of additivity of all sectional forces and in a homogeneous shear field
one obtains the following distribution of axial stress 1/1 within a chain of length L:
(5.40)
The largest stresses are encountered in the middle (z = 0) and amount to
(5.41)
Equation (5.41) predicts that for a given system (~o, Lmon, Mo) axial stresses
are proportional to shear strain and square of molecular weight. Chain rupture
occurs if 1/1max reaches or exceeds the chain strength 1/1b resulting in a gradual
decrease of M. The decrease of the average molecular weight diminishes if M ap-
proaches the limiting value M(oo) which is derived from Eq. (5.41) for the chain
strength 1/1b. The Eqs. (5.41), (5.38), and (5.6) together predict that [1]] € M(oo) is
constant for a given polymer/solvent system. Kadim [26] obtained for a 0.5 to
1% solution of PIB in decalin [1]] 7 M(oo) = 1.34 kJ/mol. He employed shear rates
between 25 and 700 S-I. Abdel-Alim et al. [27] report on a relation of 7°·41 M(oo) =
= const. for a 2% solution of polyacrylamide in water subjected to shear rates of
between 3000 and 100000 S-I.
Within more concentrated solutions the physical-chemical behavior of long
chains (size and interpenetration of molecular coils, kinetics of entanglement forma-
tion) seems to be much more important than the rheological behavior characterized
°
by ~ and €. A striking demonstration of this fact was given by Breitenbach, Rigler
and Wolf [28], who prepared solutions of 3.6 to 14.2% by weight of polystyrene
in cyclohexane. These systems showed a concentration-dependent phase separation
at a temperature Ttr between 26.4 and 29.4 °C. Shearing these solutions at a rate
of 600 S-1 and at temperatures slightly above Ttr they observed a dramatic increase
in the rate of degradation when approaching Ttr . At Ttr + 11.6 K no noticeable
109
degradation occurred within 20 h. At Tr + 0.6 K a 13% decrease of the limiting vis-

cosity [11] was measured after one hour already. Within 20 h a reduction of the molec-
ular weight from ~ 7 . 10 5 g/mol to 1.6 . 105 could be achieved.
A related subject of growing interest is the ultrasonic degradation of polymers
in solution. In their recent review Basedow and Ebert [29] summarize that ultrasonic
chain scission is the result of cavitation, i.e. of the nucleation, growth and collapse
of (gas-filled) bubbles. Chain scission occurs in the convergent flow fields in the vi-
cinity of collapsing bubbles and through the ensuing shock waves. The properties of
the solvent do not seem to be of particular importance, but the presence of nuclei
to initiate cavitation and a minimum intensity of the ultrasonic field (environ 4 W/cm 2 )
are necessary conditions. Owing to the nature of the degradation process, i.e. to the
shear loading of chains, the rates of degradation decrease with decreasing molecular
weight (degree of polymerization) and become infinitely small at a so-called limiting
molecular weight. For polystyrene in tetrahydrofuran degraded 88 h at 20 kHz
Basedow and Ebert cite one limiting molecular weight of 24000 g/mol and a newer
value of 15000 g/mol. In a quite recent publication Sheth et al. [39] report for the
same system the appearance of appreciable amounts of polymer material at molecular
weight of 1000 g/mol or lower in the gel permeation curve of the degraded sample.
They indicate that the limiting degree of polymerization depends on the initial
molecular weight distribution and is either much lower than previously reported
or perhaps non-existent.
The stresses encountered by a chain displaced with respect to a solid matrix can
also be described utilizing the concept of the monomeric friction coefficient ~o [25].
The implications of such an assumption are discussed in detail by Ferry [25] who
also lists the numerical values of the monomeric friction coefficients for many high
polymers. Naturally the coefficients strongly depend on temperature. But even if
compared at corresponding temperature, e.g. at the individual glass transition tem-
peratures, the monomeric friction coefficients vary with the physical and chemical
chain structure by 10 orders of magnitude. In the upper region one finds the
1740 Ns/m for Polymethylacrylate, the 19.5 Ns/m for Polyvinyl acetate, and the
11.2 Ns/m for Polyvinyl chloride, each at the respective glass transition temperature
[25]. This means that a PVC segment pulled at 80 °cthrough a PVC matrix at a
rate of 0.005 nm/s encounters a shearing force of 0.056 nN per monomeric unit.
At lower temperatures the molecular friction coefficient increases essentially in pro-
portion to the intensity of the relaxation time spectrum H (T), the increase being
about one to two orders of magnitude with every 10 K temperature difference [25].
If a PVC segment has a length of 2 L = 10 nm and is part of a PVC matrix strained
at a rate of 0.001 / s at the glass transition temperature it experiences at the center
an axial force of magnitude 1.1 nN. This force is about as large as that statically
transferred upon extended PE chains by PE crystallites at room temperature. It
may be repeated that the above estimate on the level of axial forces is based on the
assumption that in sample straining an extended chain segment is subjected to shear-
ing forces determined by the monomeric friction coefficients ~ 0 and that the seg-
ment does not relieve axial stresses through conformational changes. Using the same
assumptions one expects forces up to 206 nN for PMMA and up to 0.01 nN for PS
each at the respective glass transition temperature and a strain rate of O.OOI/s.
110
III. Deexcitation of Chains
This estimate shows that in dynamic straining of glassy polymers very large
axial forces can be transferred onto extended chain segments if the straining occurs
at a temperature not higher than the glass transition temperature (PMMA) or some
10K (PVC) to 20 K (PS) lower than that. These forces are sufficient to cause chain
breakage.
A loading mechanism not to be investigated in detail in the context of this mono-
graph is the straining of chain molecules through forces of inertia, e.g. through prop-
agating stress waves. Brittle thermoplastic materials (pS, SAN, PMMA) behave
"classically" at velocities of uniaxial straining below 3 mls or strain rates below
50 S-1 [30]. In this region increase the strength properties and decrease the elonga-
tion with increasing strain rate. At strain rates between 50 and 66 S-1 a transition
to stress wave-initiated fracture was observed which was accompanied by a tenfold
reduction of the apparent load-carrying capability [30]. Skelton et al. [40] studied
PA 6, PETP, and aromatic polyamide (Nomex). At ambient temperatures and in
the range ofloading rates between 0.01 and 140 S-1 these fibers also behaved clas-
sically. At -67°C and -196°C a decrease in strength with loading rate was observed
at a loading rate of 30 S-I.
The mechanisms of deexcitation of stressed chains have been discussed en passant

while investigating bond strength and chain loading. These mechanisms are chain
slip with respect to a surrounding matrix (enthalpy relaxation), change of chain con-
formation (entropy relaxation), or chain rupture.
The retraction of a stressed chain through slippage leads to an exponential stress
relaxation
IjI = ljIo exp (-tlr). (5.42)
If the slippage is adequately described by the monomeric friction coefficient

ro the relaxation time r will be given by:
r = ro Lmon/ E q (5.43)
where q is the chain cross-section. For a PVC chain at the glass transition temperature
one obtains a relaxation time of 0.05 s.
The effect of a change of conformation on the elastic chain energy had been
demonstrated in Figure 5.1. The annihilation of 4 gauche conformations within a
PE segment of 5 nm length corresponds to an increase in length of the segment of
0.25 nm. This increase of 5% will reduce the axial elastic forces acting on the chain
by 0.05 E, i.e. by::::: 10 GN/m 2 . In static straining this will correspond to complete
unloading since the largest elastically transmitted forces yielded axial stresses of
only up to 7.5 GN/m 2 . The stress-biased change of chain conformation will lead,
therefore, to a considerable reduction of axial stresses. The rate of conformational
111
changes - even of high Tg polymers and in a stress-free state - is sufficiently large

and it will be further increased under stress [19, 31].
Conformational changes under stress have been observed repeatedly (e.g. 41-45).
Gafurov and Nowak [41] studied the IR absorption of PE fibers in the range of
1200 to 1400 cm- 1 • The extinctions offour bands (at 1370, 1350, 1305, and
1270 cm- 1 ) observed there and related to the (wagging) vibra1ions of methylene
groups in gauche conformations decreased with strain. The authors estimated that
at 4% sample strain (which corresponded to much larger amorphous strains) the
number of gauche isomers in the amorphous regions was decreased by 20%.
Zhurkovet al. [42] derived the degree of chain orientation from the dichroism
of "deformed-IR bands" of PETP (cf. Chapter 8 I B and Fig. 8.5); they observed
that for highly stressed segments an initial orientation with cos 2 e = 0.75 had turned
into practically complete orientation (cos 2 e ~ 1.0). The problems posed by an ap-
plication of this method are discussed in detail by Read et al. [43]. Bouriot [44]
deduced from IR measurements a reversible cis-trans transformation of the ethylene
glycol segment in PETP fibers under the effect of tensile stresses. In polyamide 66
he observed a reversible increase of free (non-associated) NH groups and an increase
in the average chain orientation with chain tension.
Conformational changes due to straining of highly uniaxially oriented PETP
films have been reported by Sikka et aI. [45]. These changes have been inferred from
a comparison of the IR spectrum of unloaded and loaded PETP films. Employing a
highly sensitive Fourier Transform IR spectrometer Sikka noticed frequency shifts
and splittings of the absorption bands in the subtraction spectra (stressed PETP IR
spectrum minus unstressed PETP IR spectrum). At a stress of the order of 20% of
the breaking stress those IR bands were affected which are assigned to the gauche
conformations of the C-O bond next to the C-C bond. These conformations are
most likely to exist in the amorphous regions of the highly uniaxially oriented semi-
crystalline polymer. The stress-induced conformational changes seemed to include
all angles of bond rotation from minute ones to the completed gauche-trans transi-
tion.
The third mechanism of releasing axial stresses is the breakage of a chain through
homolytic bond scission, ionic decomposition or degrading chemical reactions.
It is assumed that chain scission is a thermo-mechanically activated process. A
chain segment containing ne weakest bonds of energy Uo, and carrying at its ends
a constant stress 1/10 will, on the average, break after time Te. One refers to I/I e as the
time-dependent strength of a chain segment:
(5.44)
The quantities Wo and {3 have been defined as bond vibration frequency and
activation volume for bond scission respectively [32]. AS nm PA 6 segment, for
instance, at room temperature, has an instantaneous strength of 20 GN/m 2 (taking
Uo = 188 kJ/mol, Wo = 10 12 S-1, ne = 12, Te = 4 s, (3 = 5.53 . 10- 6 m 3 /mol). If in
straining a segment, a stress of 20 GN/m2 is maintained for a period of 4 s the seg-
ment has a probability of failure of 1 - lie = 63%. In view of the different interpre-
tations found in the literature concerning the significance of wo, {3, and even Uo the
assumptions made in deriving Eq. (5.44) will briefly be discussed.
112
A chain in thermal contact with its surrounding can be represented by a system

of coupled oscillators. The degree of excitation of individual oscillators (modes of
vibration) is statistically changing. In the absence of external mechanical forces
chain scission at a C-C bond will occur whenever an oscillator representing a C-C
stretch vibration becomes excited beyond a critical level, Vo, the "strength" of the
C-C bond. In Figure 4.1 the potential energy scheme was given illustrating different
states of vibrational energy, the bond strength V o , and the dissociation energy D.
One is interested in the rate of dissociation events of oscillators activated beyond
Vo·
To evaluate this problem it will be assumed that Boltzmann statistics can be
applied to the system of, say, s oscillators representing the C-C stretch vibrations
within a volume element of the polymer. The fraction of energy states in which this
system s will have a total energy level 1;. = rE is proportional to the degeneracy gr
of the energy level Er and the Boltzmann factor exp (- Er/RT). (E is the - constant
spacing between oscillator energy levels). The partition function, therefore, will be
given by:
Qt = ~ gr exp (-Er/RT). (5.45)

r
The number of states in which at least one oscillator has an energy E j > Vo is
given by Qa :
Qa = ~ g; exp [-(Er + Vo)/RT]. (5.46)

r
For thermodynamic equilibrium within the volume element comprising s os-

cillators the fraction K+ of activated states, i.e. states with E j > V o , is given by:
(5.47)
This can be approximated [33] if r ~ 1 by:
(5.48)
where gm is the degeneracy of the most probably energy level. Expressing gm by

the entropy S according to:
S= RQn gm (5.49)
one obtains
K+ = exp (-Vo/RT) exp I:!.. SIR (5.50)
which is identical with the relation Qn K+ =- I:!.. F+ IRT. The rate of dissociation,
kb' then follows from K+ [341:
kb = K kT
h K
+
= K kT
h exp (-Vo/RT) exp I:!..SjR (5.51 )
113
s. Mechanical Excitation and Scission of a Chain
where" is the transmission constant which indicates how many of the activated
complexes actually disintegrate. By combining all front factors and exp ~S/R one
may rewrite this equation as:
kb = Wo exp (-Uo/RT). (5.52)
The quantity Wo is proportional to T; this dependency is generally negligible

against the T-dependency caused by the exponential term.
The rate kc for the breakage of one out of ne equal and independent bonds is
ne times the rate kb for the breakage of an individual bond. If the bonds are not
independent it will mean that more than one oscillator has to acquire enough vi-
brational energy for dissociation. Depending on the coupling such a cooperative
region may count some 5 or more oscillators [35J. In this case an experimentally
determined term Uo stands for the total vibrational energy required by all (coupled)
atoms which are involved in one chain rupture event. This could mean that the poten-
tial barrier between neighboring sites of an uncoupled atom is much less than Uo .
This meaning of Uo has to be kept in mind if the numerical value Uo is being com-
pared with other activation energies.
Another problem in connection with Uo arises from the fact that in a kinetic
experiment such as the determination of the rate of chain scission one derives the
bond strength Uo, that is the maximum of the potential barrier (cf. Fig. 4.1). In
calorimetric measurements the reaction enthalpy (dissociation energy D) is deter-
mined.
By application of external forces the free energy of the activated state is changed
with respect to the ground state. One may say that ~ F+ has to be replaced by
~ p+ + ~ F ext. The term ~ F ext will be some function f of the stress t/I which is act-
ing on the molecular chain along its axis; Equation (5.52) then becomes:
k =n w exp _ Uo + f(t/I) (5.53)

ceO RT·
The form of f is subject to discussion [35-37J. Linear and quadratic relation-

ships have been proposed. A linear relationship follows if one assumes that the force
acting on a particular chain is not changed by the displacement of the chain under
action of the force (comparable to a dead weight loading). The correction term f (t/I)
will be quadratic in t/I if the stored elastic energy within a small chain section to
either side of the point of chain scission is thought to be responsible for the rate
of chain fracture. From the Morse potential one obtains a difference of square root
expressions which, however, results in an almost linear relationship between force and
activation energy in the force region of interest [36J. The effect of a linear stress
potential on the intramolecular binding energy is shown in Figure 5.9.
Some information concerning the form of f( t/I) and the relation between molec-
ular stress t/I and macroscopic stress a has been obtained by Zhurkov, Vettegren
et al. [37, 32J. These authors have studied the effect of macroscopic stress a on in-
frared absorption of oriented polymers. The frequencies of skeleton vibrations of
polymer molecules show the effect of external axial forces on molecular binding
force-constants. The shift and change of shape of suitable infrared absorption bands
114
III. Deexcitation, of Chains
40
\
i \ ~
kenl
mol
20 i \
c--\---l
U
~ 'r- n
\ ~m
o i . /,'-'-:-",::::,,-::,:--- ---
i \\.i
~
Ii\
Do UO
-20
. i\
-40
\~
\
i
~---- ------ ---
-60
o 0.2 0.4
Fig. 5.9. Effect of axial stress on intra-
0.6 nm 0.8
[ - molecular binding potential (from [19 D·
should reveal, therefore, the intensity and distribution of actual molecular stresses
1/1. Experiments carried out with polypropylene and polyamide 6 have shown that
for a stressed sample the maximum of the investigated absorption band decreases
slightly and shifts towards lower frequencies [37, 32]. In addition a low frequency
tail of the absorption band appears. In agreement with existing structural models
the authors interpreted their results in assigning the shifted symmetrical part of the
absorption band to the crystalline phase of the sample, the low frequency tail to
the amorphous phase. The shift of the symmetrical part of the absorption band,
b. vs , turned out to be linear with macroscopic stress in all performed experiments:
(5.54)
Following a theory of Bernstein [38] the shift of the stretching frequency of

a bond can be related to a shift in dissociation energy D of the same bond:
(5.55)
Since b. Vs ~ Vo one may substitute v5 - v~ by b. Vs . 2 vo. A consequence of

this condition is that also (v5 - v~) is proportional to a. The choice of v or v2 in
Eq. (5.54) only affects a, not the power of a. If one further assumes that the activa-
tion energy Uo is changed by a similar amount as Do one obtains:
2vo
Ua = Uo - - aa. (5.56)
c
115
In order to derive the form of f( t/I) the relation between u and t/I for oriented
crystalline phases has to be established. According to experimental and theoretical
evidence it is assumed that a condition of homogeneous stress exists within the
crystalline layers of a regular sandwich structure and that t/I is proportional to u.
Equation (5.56) thus yields that the activation energy for scission of an oriented
chain molecule is linearly decreased by stresses acting upon this molecule. Equation
(5.53) with (5.56) may be written as:
ke = ne Wo exp [(-Vo + 13 t/I)/RT] (5.57)
which is equivalent to Eq. (5.44). It is believed that the rate of fracture of chains
all subjected to the same local stress t/I can correctly be described by Eq. (5.57).
On the basis of the foregoing considerations the response of two chains of dif-
ferent conformation to axial stresses has been evaluated in Figure 5.10. An extended
chain segment (L = Lo) stretches elastically up to the time-dependent fracture stress.
The calculations have been carried out employing a modulus of elasticity for the
polyamide 6 chain of 200 GN/m2 • As indicated in Table 5.3 it is still open to dis-
cussion whether this value can be unambiguously accepted. The times to fracture
have been calculated from Eq. (5.57) for a segment comprising 12 bonds of low
strength (Vo = 188 kJ /mol). The partially extended chain (L = 1.1 Lo) contains
initially four kinks (8 gauche bonds). It responds in region A (Fig. 5.10) mostly by
the hindered rotation of its gauche bonds, which, in this region, accounts for 90%
of the axial deformation. The resulting combined modulus amounts to 22 GN/m2 .
If geometrically possible a transition to the fully extended conformation will occur
in region B. Region C represents the energy elastic straining of the then extended
chain segment of length 1.1 Lo.
The degradation of chains through ionic decomposition or chemical reactions
under stress will be discussed where appropriate in the following chapters.
1jl
25
GN/m2
20
15
10
o 5 10 15 % 20
Fig. 5.10. Stress-strain diagram of polyamide 6 chain segments. L = contour length, 5.2 nm,
12 weak bonds, LO = distance between segment ends in the unstressed state, T e = like = average
lifetime according to Eq. (5.57), A =response due to hindered rotation of gauche bonds, B =
intermediate region, C = energy-elastic response of extended chain.
116
1. P. J. Flory: Principles of Polymer Chemistry, Ithaca, New York: Cornell University Press
1953.
2. 1. R. G. Treloar: The Physics of Rubber Elasticity, 2nd Ed., London: Oxford University
Press 1958, a) p. 61, b) p. 103ff, c) p. 108.
3. H. M. James, E. Guth: J. Chern. Phys. 11,455 (1943).
4. A. Abe, R. 1. Jernigan, P. J. Flory: J. Am. Chern. Soc. 88,631 (1966).
5. J. P. Flory, A. D. Williams: J. Polymer Sci. A-2, 5,399 (1967).
6. P. R. Saunders: J. Polymer Sci. A-2, 2, 3765 (1964).
7. W. Pechhold: Kolloid-Z. Z. Polymere 228, 1 (1968).
8. H. G. Zachrnann: Kolloid-Z. Z. Polymere 231, 504 (1969).
9. J. A. Sauer, A. E. Woodward: Polymer Thermal Analysis, II., P. E. Slade, Jr., and L. T.
Jenkins, Eds., New York: Dekker, 1970, p. 107.
10. P. B. Bowden: Polymer 9, 449~454 (1968).
11. D. S. Boudreaux: J. Polymer Sci.: Polymer Phys. Ed. 11, 1285 (1973).
12. W. J. Dulmage, L. E. Contois: J. Polymer Sci. 28,275 (1958).
13. T. R. Manley, C. G. Martin: Polymer 14, 491~496 (1973) and ibid. 632~638 (1973).
14. H. Hahn, D. Richter: Colloid + Polymer Sci. 255. 11l~-119 (977).
15. E.g. J. Meixner: Handbuch der Physik, Vol.lII/2, S. Fliigge, Ed., Berlin: Springer-Verlag
1959, p. 413.
16. D. Langbein: Theory of Van der Waal's Attraction, Springer Tracts in Modern Physics,
Vol. 72, Heidelberg: Springer-Verlag, 1974.
17. P. Lindenmeyer: Mechanical Behavior of Materials, Kyoto, Japan: Soc. Mat. Sci. 1972,
Special Vol., p. 74.
18. H. Scherr: Dissertation Universitiit Ulm, J uli 1973.
19. H. H. Kausch: J. Polymer Sci. C 32 (Polymer Symposia), 1 (1971).
20. A. D. Chevychelov: Polymer Science USSR 8, 49 (1966).
21. H. H. Kausch, D. Langbein: J. Polymer Sci., Polymer Physics, Ed. 11, 1201 ~1218
(1973).
22. H. H. Kausch, J. Becht: Deformation and Fracture of High Polymers, H. H. Kausch
et aI., Eds., New York: Plenum Press 1973, p. 317.
23. H. Baur, B. Wunderlich: Fortschr. Hochpolymeren Forsch. 7, 388 (1966).
24. E. W. Fischer, G. Hinrichsen: Kolloid-Z. Z. Polymere 213,28 (1966).
25. J. D. Ferry: Viscoelastic Properties of Polymers, 2nd Edition, New York: J. Wiley and Sons,
Inc. 1970.
26. A. Kadim: Thesis, Dept. of Chern. Engng., Technion, Israel Institute of Technology, Haifa,
Israel 1968.
27. A. H. Abdel-Alim, A. E. Hamielec: 1. Appl. Polymer Sci. 17, 3769~3778 (1973).
28.1. W. Breitenbach, 1. K. Rigler, B. A. Wolf: Makromol. Chemie 164, 353~355 (1973).
29. A. M. Basedow, K. Ebert: Adv. Polymer Sci. (Fortschr. Hochpolymerenforsch.) 22, 83~143
(1977).
30. E. M. Hagerman, C. C. Mentzer: J. Appl. Polymer Sci. 19, 1507~1520 (1975).
31. Yu. Ya. Gotlib, A. A. Darinskii: Vysokomol. soyed.A 16, 2296~2302 (1974), Polymer
Science USSR 16, 2666 (1974).
32. S. N. Zhurkov, V. E. Korsukov: J. Polym. Sci., Polymer Physics Ed. 12, 385~ 398 (1974).
33. E. A. Guggenheim: Handbuch der Physik, S. Fliigge, Ed., Vol. 1II/2. Berlin: Springer-Verlag
1959.
34. S. Glasstone, K. 1. Laidler, H. Eyring: The Theory of Rate Processes, New York:
McGraw-Hill 1941.
35. H. H. Kausch: J. Macromol. Sci., Revs. Macromol. Chern., C4 (2), 243~280 (1970).
36. E. E. Tornashevskii: Soviet Physics ~ Solid State 12, 2588~2592 (1971).
37. S. N. Zhurkov, V. 1. Vettegren', V. E. Korsukov: 2nd Internat. Conf. Fracture, Brighton,
England April 1969, paper 47.
117
38. H. I. Bernstein: Spectrochim. Acta 18, 161 (1962).

39. P. J. Sheth, J. F. Johnson, R. S. Porter: Polymer 18, 741-742 (1977), C. B. Wu, P. J.
Sheth, J. F. Iohnson: Polymer 18, 822-824 (1977).
40. J. Skelton, W. D. Freeston, Ir., H. K. Ford: Appl. Polymer Symposia 12, 111-135 (1969).
41. u. G. Gafurov, I. I. Novak: Polymer Mechanics 6/1, 160-161 (1972).
42. s. N. Zhurkov, V. I. Vettegren', V. E. Korsukov, 1.1. Novak: Soviet Physics - Solid State
11 /2, 233-237 (1969).
43. B. E. Read, D. A. Hughes,D. C. Barnes, F. W. M. Drury: Polymer 13, 485-494 (1972).
44. P. Bouriot: Premier Colloque du Groupe Frans:ais de Physique des Polymeres, Strasbourg,
21-22 mai 1970.
45. S. Sikka: PhD Thesis: Application of FTIR to study external stress effects on the PETP
backbone, University of Utah, Salt Lake City, December 1976. S. Sikka, K. Knutson: to
be published.
46. V. S. Kuksenko, V. A. Ovchinnikov, A.1. Slutkser: Vysokomol. soyed All 9,1953-1957
(1969).
47. H. Hahn: Personal communication.
118
Chapter 6
Identification of ESR Spectra of

Mechanically Formed Free Radicals
I. Formation . 119
II. EPR Technique 120
A. Principles 120
B. Hyperfine Structure of ESR Spectra 121
C. Number of Spins 122
III. Reactions and Means of Identification . 123
IV. Assignment of Spectra. 125
A. Free Radicals in Ground High Polymers 125
B. Free Radicals in Tensile Specimens . 126
I. Formation
The action of axial tensile forces on a molecular bond RI-R2 results in a decrease
of the apparent binding energy of the bond and thus in an increase of the probability
for bond scission. If the reduction of the apparent binding energy is sizeable the
mechanical action may be considered as the main cause of chain destruction. In as
much as the scission of a chain molecule into organic radicals and the resulting ap-
pearance of unpaired free electrons is governed by mechanical forces a study of
radical formation and of radical reactions will reveal information on the forces acting
on a chain at a molecular level. The method of investigating free radicals by para-
magnetic resonance techniques has been highly developed during the last thirty years
[1, 2]. Since then it was successfully applied to elucidate the mechanism of the for-
mation of free radicals in chemical reactions and under irradiation of visible and
ultraviolet light, of x- and ,),-rays, and of particles [1.3]. Also the value of the spec-
troscopic splitting factor g, the magnetic surrounding of the unpaired free electron
spin, and the structure of the free radical have been studied. In all these cases the free
electron spin acts as a probe, which, at least temporarily, is attached to a certain
molecule, takes part in the motion of the molecule, and interacts with the surround-
ing magnetic field.
In investigations of the deformation and fracture of high polymer solids [4-67] by
electron paramagnetic resonance techniques (EPR) similar experimental problems
arise as during the above mentioned applications of EPR:
1. Interpretation of the form of the observed spectra, i.e. determination of the
position of the free electron within the molecule.
119
6. Identification of ESR Spectra of Mechanically Formed Free Radicals
2. Measurement of spectrum intensity, i.e. determination of the number of free

electrons.
3. Observation of the changes displayed by the EPR signal and elucidation of the
nature and kinetics of radical reactions.
From these primary data, information is sought to obtain with respect to the
deformation behavior of chain molecules, their scission and their role in fracture
initiation.
In the following a brief description of EPR spectroscopy will be given in order
to provide an understanding of the main experimental technique which is used in
these investigations. Subsequently reactions, identification, and concentration of
mechanically formed free radicals will be discussed.
II. EPR Technique
A. Principles
The EPR spectroscopy of free electrons is based on the paramagnetism of free electron
spins. For this reason one also speaks of electron spin resonance (ESR) spectroscopy.
Electrons on fully occupied molecular orbitals generally do not give rise to a magnetic
moment since - according to the Pauli principle - pairwise spins compensate each
other. If, however, a bond is broken through homolytic scission, then free radicals
with impaired electron spins are formed which are detectable. A free electron has the
magnetic moment IJ. given by
(6.1)
where g is the spectroscopic splitting factor, {3 is the Bohr magneton, and s is the
spin vector. In a magnetic field H the spin may assume one of two positions (spin
quantum numbers m being either + 1/2 or - 1/2). The energy difference between
these two states is
AE = g{3H. (6.2)
The populations of the two energy levels at equilibrium are given by Boltzmann
statistics which indicate that more of the free electrons will normally reside in the
- 1/2 "ground state". However, when electromagnetic energy of an appropriate
(microwave) frequency v is supplied, an electron can be "pumped" to the higher
+ 1/2 state. The resonance condition is fulfilled when hv is just equal to the energy
difference between the two levels:
AE = hv = g{3Hres (6.3)
where h is Planck's constant. In resonance, microwave energy is absorbed by the

resonating sample if the upper energy level is constantly depopulated by thermal
interaction between the spins and the molecular lattice. Generally the microwave
120
II. EPR Technique
frequency employed by commercially available ESR spectrometers is fixed (at about

9.5 GHz) and the magnetic field is swept through the resonance (at about 3200 Oe).
B. Hyperfine Structure of ESR Spectra
Since the "free" electrons are attached to molecules, they will sense not only the
presence of the applied external magnetic field but will show magnetic interactions
with nuclei of surrounding atoms which have a magnetic moment. These interactions
give rise to the "splitting" that can be used to identify particular radicals. Equa-
tion (6.3) is accordingly transformed into
n
Hres = h ZJ / g ~ - L (Ai + Bi) mIi. (6.4)
i= 1
Here mn is the spin quantum number of the interacting nucleus i, Ai the iso-
tropic part of the magnetic coupling between radical electron and nucleus i, and
Bi the anisotropic part which depends on the orientation of the radical within the
external magnetic field.
For the important class of hydrocarbon radicals, B is different from zero only for
the a-protons, i.e. for the nuclei of hydrogen atoms directly bound to the radical
C atom. In well ordered systems (highly oriented polymers, single crystals) these
radicals do have a strongly orientation dependent hyperfine structure.
In the simplest case of no interaction (A = 0, B =0) there will be one line with
its center at Hres . If only one proton interacts isotropically (B =0) with the radical
electron then the spectrum will show two lines displaced from Hres by ±A. If two
equivalent nuclei interact with the electron then m! can assume the values + 1, 0,
and -1 so that a triplet of lines appears with distances of -A, 0, and +A from the
center respectively. Since the state with m! = 0 appears twice as often as the other
ones the individual resonance lines have intensitfes as 1: 2: 1. It may generally be said
that n equivalent protons give rise to a spectrum with (n + 1) lines which show an
intensity distribution given by the binomial coefficients of exponent n. For illustra-
tion Figure 6.1 shows the spectra of alkyl radicals -CH 2 -CH-CH 2 - in highly
{3 {3
oriented polyethylene monofilaments which were produced by irradiation with fast
electrons [37]. In one orientation (fiber axis parallel to H, spectrum at bottom) the
total splitting parameter (A + B) of the a-protons is just equal to the isotropic para-
meter of the 4 equivalent {3-protons. Therefore, a spectrum is observed which might
be caused by 5 equivalent protons, i.e. a sextet with intensities 1: 5: 10: 10: 5: 1 (some
background including the small center line belongs to a second radical present within
the sample). In the case of a perpendicular arrangement of fiber axis and magnetic
field the spectrum shown in the upper portion of Figure 6.1 is observed. Now the
a- and ~-protons have different splitting parameters and a quintet of lines appears
due to 4 equivalent ~-protons with each quintet line split into a doublet by the
a-proton. In all 10 lines result with intensity ratios of 1:1:4:4:6:6:4:4:1:1:. Again
the small background of a superimposed second spectrum is observed. The orienta-
121
500. - H
500. - H
Fig. 6.1. Orientation dependency of the hyperfine structure of

main chain (alkyl) free radicals: a. axis of highly oriented poly-
ethylene fiber parallel to magnetic field Ho, b. axis of fiber per-
pendicular to Ho (after [37,38]).
tion dependency of the hyperfine structure of hydrocarbon radicals offers the op-
portunity to study the orientation of radical carrying chains.
C. Number of Spins
In most spectrometers lines are recorded which are the first derivative of the energy
absorption peaks with respect to the magnetic field Ho. It is the plot of these first
derivates that is usually called the ESR spectrum.
The total microwave energy absorption can be obtained by double integration
of the ESR spectrum. In the absence of saturation effects the total power absorbed,
1(00), depends upon the number of spins No, the intensity of the oscillating magnetic
(microwave) field HI' and the frequency v according to
1(00) = 21T2 v2 H2 _1_ N g2 {32. (6.5)

14kT 0
To calculate absolute intensities of spins usually comparison of the unknown

intensity No is made with a standard sample with a known concentration Ns of spins
[I]. Frequently a solution of the stable free radicall,l-diphenyl-2-picryl hydrazyl
(DPPH) in benzene is used as standard radical. If the "unknown" spectrum is sym-
122
III. Reactions and Means of Identification
metrical and narrow and the experimental conditions (sample size and shape, con-
ditions of spectrometer operation) are the same in both cases the relation
(6.6)
is valid where Ns and Is (00) refer to the standard sample. If lines of identical shape
and width are compared it is sufficient to measure the peak height h of the first
derivative. The numbers N 1 and N2 of unpaired electrons causing these peak heights
are then related by:
(6.7)
For most cases of stress induced free radical formation this relation has been
used for evaluation of time-dependent intensities.
The number of free radicals necessary for detectable peak heights depends on
the line shape and the effective volume of the spectrometer cavity. At the time
covered by the referenced investigations (1960-1976) the lower limit of detectabili-
ty of polymer radicals in the solid state can be given as 10 13 spins/ cm3 .
It may be mentioned that the number of free radicals present in a liquid or at
the surface of a solid can also be determined by a non-spectroscopic method. This
method exploits the color changes associated with the consumption of free radicals
from a test solution e.g. of DPPH [39]. It should be emphasized again that great
caution must be exercised in comparing absolute values of free radical concentra-
tions obtained from differently shaped samples and/or in different cavities. Accord-
ing to general experience deviations of up to a factor of 2 must be expected.
III. Reactions and Means of Identification
The assignment of the specific peaks to specific types of free radicals is not always
obvious; it sometimes involves highly intriguing techniques and the knowledge of
related spectra and of chemical reactions to be expected [64, 67]. The main obstacles
to be overcome are the broad line widths of resonance lines from solid specimens
and the high rate of many radical reactions. It is evident that a large line width often
prevents an effective resolution of nuclear hyperfine structure. The so-called
"5 + 4 line" spectrum of mechanically destroyed methacrylic polymers [4] is an
example of a spectrum which had only been identified after comparison with the
16-line-spectrum of an aqueous solution of the polymerization radical of methacrylic
acid. Thus it was learned that the former spectrum is the unresolved form of the latter
and has to be assigned to the same radical [40].
The difficulties in identifying an observed spectrum are well illustrated by the
spectrum of 6 polyamide shown in Figure 6.2. It was obtained after stretching a fiber
sample at room temperature in the ESR cavity by an automatic servo-control system
to the point of radical detection, then scanning the magnetic field from 3071 to
3571 Oersted (mean, 3321 Oe) at 9.433 GHz and a time constant of 0.5 second.
123
Fig. 6.2. Apparent 5-line spectrum

as obtained after stretching of a
bundle of 6 polyamide fibers at
room temperature in an ESR cavity.
This apparent 5-line spectrum is obtained, under exclusion of oxygen, when highly
stressed fibers or finely ground or milled material are investigated at room tempera-
ture [41]. A similar spectrum is observed after 'Y or electron irradiation of PA 6
[42, 43). Two different interpretations of this 5-line spectrum have been given. Graves
and Ormerod [42] suggested that an apparent quartet coming from the radical
-CH2-CO-NH-CH-CH2- (radical I) is superimposed by a singlet which may be
due to -CH2-CHO-NH-CH2- (II). Kashiwagi [43] deduced, after comparison with
radicals from polyurea and succinamide, that the apparent 5-line spectrum from
6 polyamide and the 6-line spectra from 610, 66 and 57 polyamides are a triplet of
doublets stemming from the radical!. It is, therefore, generally concluded that the
radicals observed at room temperature in mechanically degraded polyamide material
are mainly of type I, superimposed to a varying degree by a component of type II.
The radicals I are secondary chain radicals which can only have been produced by
the reaction of a hydrogen from a methylene group with an other - a primary -
radical. Since the secondary radicals are reaction products they give no direct in-
formation on the nature and location of the primary process of mechanical degrada-
tion of the above solids.
The appearance of free radicals in the process of mechanical degradation reveals
conclusively that under the action of local mechanical forces main chains have been
broken and that the breakage of a main chain bond has led to the formation of a
chain end free radical at either side of the broken bond. Knowledge of these facts
can already be employed to study the relation between environmental parameters
and the rate of chain scission. The kinetics of chain scission and its possible role in
fracture initiation is investigated in detail in Chapters 7 and 8. Apart from the inten-
sity which gives a quantitative measure of chain scission the form of free radical
spectra has been used to determine the location of primary and secondary radicals
within the molecule [64] and within morphological units (within crystals, on crys-
tal surfaces, in amorphous regions).
If degradation and ESR investigation of a sample are carried out at liquid nitrogen
temperature, the rate of radical reactions is slowed down sufficiently and a direct
observation of the primary radicals produced by mechanical degradation becomes
124
IV. Assignment of Spectra
possible. In a thorough investigation Zakrevskii, Tomashevskii, and Baptizmanskii

[10] have cleared up the scheme of radical reactions for 6 polyamide (caprolactam,
capron). At 77 K they observed a complex spectrum showing the hyper fine struc-
ture of a sextet superimposed on a triplet. From the distance between the different
lines of the sextet (splitting) and their intensity ratios the above authors established
the presence of the radical R-CH2 -CH2 (III). This radical will form after rupture
of any of the bonds 1 to 6 in the caprolactam unit:
If the superimposed sextet is subtracted from the original spectrum a triplet

remains which must be due to a scission product of the above molecule. The ex-
clusive breakage of bonds 3 and 4 can be ruled out immediately since it would only
lead to radicals of type III. Breakage of bonds 6 and 7 was eliminated after compari-
son with capron deuterated in the imino groups; breakage of bond 1 would not lead
to a radical with a triplet spectrum. After inspection of all other possibilities, in-
cluding secondary radicals, there remain as likely scission points only the bonds
2 and 5. After having studied a and e methyl substituted caprolactam the authors
[10] finally were able to state that in a stressed 6 polyamide molecule both the bonds
2 and 5 do break and with equal probability. The rupture of a PA 6 molecule, there-
fore, leads to three primary radicals:
-CH2-CH2 (III, 50%), -NH-CH2 (IV, 25%) and -CO-CH2 (V, 25%).
Upon slowly rising the temperature from 77 K radical III begins to disappear
leaving at 120 K only IV and V. At higher temperatures IV and V are gradually trans-
formed into I which prevails above 240 K. Within the experimental limits mentioned
above with respect to comparing the absolute numbers of radicals of different species
there seems to be no loss of radicals during conversion of III, IV, and V into radical I.
In the following Sections and Tables the experimental conditions are reported
which were employed during mechanical production and ESR investigation of free
radicals. The assignments of primary and/or secondary radicals are given. For a dis-
cussion of the nature and kinetics of possible transfer reactions the reader is referred
to the recent and comprehensive reviews of Ranby et al. [2] and of Sohma et al. [64].
Special problems in view of polymer morphology and of the reduction of chain
strengths will be treated in Chapters 7 and 8.
I V. Assignment of Spectra
A. Free Radicals in Ground High Polymers
The appearance of macroradicals in mechanically destroyed polymers was first shown

in 1959 [4-6]. Since then, fairly systematically natural and synthetic organic ma-
terials have been investigated with respect to formation of free radicals in mechanical
125
degradation (see for instance the monograph by Ranby and Rabek [2] and the review
papers by Butyagin et al. [7], Kausch [8], aNd Sohmaet al. [64]). Owing to the original-
ly limited sensitivity of ESR-spectrometers, the first experiments have been carried
out with crushed polymers which have a high ratio of fracture surface to volume and,
therefore, a comparatively large ESR signal.
The experimental equipment is described by various authors (referenced in 2).
Grinding, milling, cutting or crushing of the polymer sample is done in vibration mills
or mortars, under inert atmosphere, in vacuo, or in liquid nitrogen. Magnetic drills
have been reported by which the shavings or grindings are produced within a sealed
glass tube immediately at the ESR-cavity. Most equipment used in these experiments
allows for a control of the gaseous environment and for temperature control from
liquid nitrogen temperature (77 K) to above room temperature.
In Table 6.1 the presently available details of sample preparation (crushing tech-
nique, temperature, environment), the treatment of the crushed sample, the tem-
peratures at which the ESR-spectra were taken, and the assignments of the observed
spectra to primary and/or secondary free radicals are listed for 35 different polymeric
materials. The common result of practically all of the reported ESR work [4-36] on
comminuted polymers is that the mechanical action leads to the severance of a main
chain bond and to the formation of chain end radicals as primary radicals. The only
exception to this rule are the substituted polydimethylsiloxanes (No. 32 to 35) where
there is a strong indication that the Si-O bond decomposes by an ionic mechanism
rather than by homolytic bond scission [36]. In no case free radicals are produced
by a mechanical stripping off of sidegroups or -atoms from the main chain. For this
to occur stresses are required which simply cannot be transmitted onto the relatively
small sidegroups present in the materials listed in Table 6.1. In fact the attempted
degradation of low molecular weight compounds (paraffins, ethanol, benzene) with
a molecular weight equal to or larger than that of those sidegroups failed, although
mechanical means identical to those used successfully for macromolecules had been
employed [13, 14,62].
B. Free Radicals in Tensile Specimens
Mechanical degradation which leads to formation of free radicals can be achieved

in various ways but generally two types of samples are being used in experiment:
powders of finely ground polymers as discussed above or uniaxially stressed fibers,
monofilaments, or strips of films. In samples of the first type all molecular fracture
processes are completed before the ESR investigation begins. These samples can
primarily be used to obtain information on the nature and absolute number of free
radicals present within the degraded polymer.
Samples of the second type, tensile specimens, are in princi pIe suitable for study-
ing the time-dependent formation of free radicals and their effect on strength and on
other macroscopic properties. It can be said a priori that the free radical intensities
of tensile specimens - be it before or after rupture - will be very much lower than
the intensities found in finely ground polymers with fracture surfaces of several
thousand cm 2 /g. Even with the increased sensitivity of presently available ESR spec-
126
trometers the investigation of stressed samples is limited to a few with favourable

properties. The only experiments leading to a successfull investigation of time-
dependent formation of free radicals were carried out with highly oriented, semi-
crystalline fibers like 6 and 66 polyamide, polyoxamides, polyethylene, polyethylene
terephthalate, or natural silk and with preoriented rubbers (natural rubber, polychlo-
roprene) near liquid nitrogen temperatures.
In Table 6.2 experimental conditions and types of radicals observed are listed.
The experimental equipment necessary for stressing fibers within the resonant
cavity has been extensively reviewed by RAnby et al. [2]. It is quite more elaborate
than that for investigating powders, although the requirements for controlled tem-
perature and atmosphere are nearly the same. Lever systems or servo-controlled
hydraulic loading systems have been reported [29, 37, 44, 60] through which pro-
grammed load or strain functions can be applied to bundles of fibers (or other tensile
specimens) in the resonant cavity. The stretching of the tensile specimens necessarily
has to be carried out in that temperature regime within which free radical spectra can
be well observed. For thermoplastic fibers this meant 200 to 320 K; for preoriented
rubbers ambient temperatures of 93 to 123 K were necessary. At these temperatures
the primary free radicals are sufficiently mobile to react rapidly with atom groups
of the same or other chain molecules, with absorbed gases, or with additives or im-
purities acting as radical traps. For this reason the observed free radicals listed in
Table 6.2 are secondary radicals - with the only exception of the highly stable pri-
mary radical of poly-p-(2-hydroxyethoxy) benzoic acid (PEOR) studied by Naga-
mura and Takayanagi [60]. The secondary radicals - if sufficiently resolved - have
been assigned to either peroxy radicals (R-O-O), main chain radicals (R-CH-R),
unknown impurities [55], or deliberately added radical traps as, for instance chloranyl
[61 ].
127
Table 6.1. Condition of Degradation and Assignment of Resulting ESR Spectra of Polymers
Degraded Polymer Sample preparation (Heat) treatment

time at temperature
process temper- environment
ature min K
1. Polyethylene PE milling 77 liqu. N2 + 10% 02

- CH2-CH2- milling 240
milling 80-100 vacuum
milling 77 helium
sawing 77 liqu.N2 none
milling 77 helium
sawing 77 liqu.N2 5 132
milling 77 helium
sawing 77 liqu. N2 5 152
5 233
milling 77 helium warming to 270
2. Polypropylene PP milling 80-100 vacuum

-CH 2 -(CH 3)CH-
milling 80-100 vacuum warming to 140
grinding 77
sawing 77 liqu. N2
ball mill 77 vacuum
ball mill 77 vacuum introducing air
3. Polyisobutylene PIB milling di- hexane, carbon

-CH 2 -(CH 3)2 C- lute solution 77 tetrachloride
cutting 77 liqu. N2 + 10% 02
milling 77 liqu. N2 + 10% 02
Poly(1-butene) PB vibrating mill 77 inert
-CH(CH2-CH3)-CH2-
iibrating mill 77 inert 200
4. Polybutadiene BR
-CH 2 -CH=CH-CH r
cis-rich sawing 77 liqu. N2
54% trans sawing 77 liqu. N2
ball mill 77 vacuum
5. Polyisoprene milling 77 liqu. N2 + 10% 02

-CH 2-CH 3C=CH-CH 2 -
6. Polystyrene PS milling di- various solvents

-CH 2-(C 6 H s )CH- lute solution 77 plus monomers
milling room vacuum
128
ESR Assignment of observed spectra to

spectra Ref.
taken at primary radicals secondary radicals
K
I
1. 77 4,9,10
240 5
77 -CH 2 -CH-CH 2 - 11
77 -CHz-CH 2 none 12
77 none 13,14
135 -CH2-¢H-CH 3 12
77 -CH2-<;H-CH 3 13, 14
160 none -CH2-<;:H-CH3 12
77 none -CH2-CH-<;H3 and
-CH 2 -HC0<;l-CH 3 or
-CH 2 -HCO<;l-CH2- 13,14
77 none -CH 2 -HCOO- CH 3 or
-CH 2-HCOO-CH 2- 13,14
77 none -CH 2 -CH=CH-CH-CH2- 12
2. 77 -CH 2 -CH 3C-CH 2 -(increasing 11
140
1 -CH2-CH3CH
with milling time)
-CH 2 -CH 3C-CH 2 -(Sextet)
11
9
77 none 13,14
77 J ROO
77 -CH2-CH3CH and
-CH 3CH-CI:I2 none 15
77 -CH 2 -CH 3CH -CH 3CH-H 2 COO 15
3
16
77
77
l -CH 2 -(CH 3h C
R}-CH-R2
none 5
77 f ROO 4
77 -CH2-CH(CH2-CH3)-CH2 and 65
77 -CHz-CH(CH2-CH3) -CH2 -C (CH 2 -CH 3)-CH 2 - 65
77 none ROO 13,14

77 none ROO 13,14
77 none R-CH-CH=CH-CH3 62
5. none ROO 4
6
77 -CH 2 -C 6H s CH reaction of R with monomer 17
room none ROO 4
129
Table 6.1: (continued)

time at temperature
ature min K
6. (continued) milling 77 liqu. N2 + O 2

cutting 77 liqu. N2 + 10% 02
shaving 77 vacuum none
shaving 77 vacuum heating in air to 273
7. Poly (l( methyistyrene milling di- various solvents

-CH2-(C6Hs)(CH3)C- lute solu- 77 plus amines
tion 77 toluene
77 toluene
8. Polytetrafluorethylene PTFE sawing 77 liqu. N2 none

- CF2- CFr milling 77 liqu.N 2 +10%02
ball mill 77 vacuum
9. PolyvinyJalcohol PVAL grinding di- 80 H 2O;D2 0

-CH 2 -OHCH- lute solution
grinding 80 vacuum
10. Polyvinylacetate PVAC shaving 77 vacuum

-CH 2-(OCOCH 3)CH- shaving 77 vacuum heat. in vac. to 300
cutting 77? liqu. N2 + 10% O2
milling di- various solvents
lute solution 77 and monomers
11. PolymethacryJate PMA cutting 77 liqu. N2 + 10% 02

-CH2-(C02CH3)CH- milling di- various solvents
lute solution 77 and monomers
12. PolymethylmethacryJate PMMA milling 77 liqu. N2 + 10% 02 heat. in air to 300

-CH2-(C02CH)(CH3)C- milling 77 liqu.N2+ 10%02
sawing 77 Iiqu. N2
shaving room vacuum
shaving 77 liqu. N2 + 10% 02
grinding 77 vacuum
grinding 77 vacuum
grinding 77 vacuum
grinding di-
lute solution 77 benzene
13. Polyacrylonitrile PAN vibro mill 77

-CH2-CNCH- 77
vibro mill air
130
IV. Assignment ot Spectra

spectra Ref.
K
77 none ROO 4
77 none ROO 5
77 -CH 2 -C 6H s CH 18, 19, 20
300 none -CH 2 -C6 H4 CH 2 6
273 none ROO 19
I
7
77 to 300 -CH 2 -(CH 3 )(C 6H S )C reaction of Rwith amines 21
77 Rl-~H-R2 16
177 none RI-CH-R2 16
8 77 none -CF 2 - CF20 <? 13,14

77 none -CF 2 -CF 2OO 4
77 and 243 CF r CF 2 none 62
9 80-170 -CH 2 -OHCH and -OHCH-CH2 none 22
80-350 -CH 2 -OHCH -CH 2 -OHC-CH 2 - 22,23
10 77 -CH2 -(OCOCH 3 )CH none 19

room none none 19
77?
I -CH r (OCOCH 3 )CH
5
77 f reaction of Rwith monomer 17,21
}
11 77 5, 9
-CH 2 -(COOCH 3 CH
77 reaction of R with monomer 17
12.room ROO 4
77 ROO 4
77 slight.ROO 13,14
77 RI-CH-R2 19
room none 19
300 -CH2 -(COOCH 3 )(CH 3 )C none 5
77 RI-CH-R2 16,62,6 3
273 no ch~mge of RI-CH-R2 16
283 RI-CH-R2 disappears 16
77 RI-CH-R2 16
13 77 I 23
195 I -CHrCNCH 23
195 ROO 24
131

time at temperatun
ature min K
14. Poly(oxymethylene) POM sawing 77 liqu. N2

~CH2~0~ grinding 80 vacuum
15. Polyethyleneoxide PEO grinding 80 vacuum

~CH2~CH2~0 grinding di-
lute solution 80 H2 0 + CH 3COOH
16. Polypropyleneoxide grinding 80 vacuum

~CH2~CH3CH~0~
17. Polyethyleneterephthalate PETP vibro mill 77

~0~CH2~CH2~0~CO~C6H4~CO~
18. Poly carbonate PC 77

~0~C6H4~C(CH3)z~C6H4~0~CO~
19. Poly(2,6-dimethyl-p-phenylene oxide) PPO 77 Iiqu. N2 + 02

~(CH3)zC6H2~0- shaving 77 vacuum
room air
20. Polycaprolactam 6-PA cutting 77 liqu. N2 + 10% O2

~CO~(CH2)5~NH~ milling 77 various (liqu.
N2 , helium)
grinding 77 vacuum
milling 77 helium
milling vacuum
milling 77
milling 77 helium heated to 295
21. Poly 0< methylcaprolactam milling 77 vacuum

~ CO~CH(CH3)~ (CH2)4~ NH~
Poly € methylcaprolactam milling 77 vacuum
~CO~(CH2)4~CH(CH3)~NH~
22. Poly(hexamethyleneadipamide) 66-PA grinding 77 liqu. N2

~NH~(CH2)6~NH~CO~(CH2)4~CO~ 77 liqu. N2 02 intro,uced
23. Polyurethane
(Solithane 113)
PU grinding
grinding
77
room
liqu. N2
quenched
I.In
~0~CO~NH~(CH2)n~ liqu. N2
24. Natural Silk milling 77 helium

milling 77 helium
milling 77 helium
132

spectra Ref.
K
14 77 - CH2 0 - CH 2.o° 13,14

88-350 -O-CH-O-or-O-CH-OH 22
15 80-230 -0-CH2 or -+ -0-CH-CH2- 22
80-200 -0-CH2 22
16 80-200 -0-CH 3CHor -+ -0-CH3C-CH2 22
17 77 R-CO-0-CH2 25
18 77 -0-C 6H4 and -OC-0-H4C6 26
19 77
77 l - (CH 3)2C6H2- 0
ROO
none
27
27
room 1 none 27
20 77 none mostly ROO 5

77 -CH2-CH2; -CO-CH2 9,10,12,
-NH-CH2 28
77 l 10,66
155 ( -CO-CH 2 ; -NH-CH2 12
195 22
113 none -CO-NH-¢H-CH2- 9,29
77 none -CO-NH-CH-CH2- and singlet 12
21 77 -CO-CH(CH 3) 10
77 -NH-CH(CH3) 10
22 77 -CO-CH2; -NH-CH2 -CO-;-NH-CH-CH2 30,31,32

233 ROO 30, 32
23. 77 singlet 30,33

77 singlet 33
24 .77 -NH-CH2; -CO-CH2 none l 12

120 -NH-CH- CH 3;-CO-CH-CH 3 -NH-CH-CO- 12
295 none ( 12
133

time at temperature
ature min K
25. Polyethylene-Sulfide milling 77 liqu. N2

-CH 2 - CH2- Sn-
26. Polypropylene-Sulfide milling 77 liqu.N2

-CH 2 -CH 3CH-S n -
27. Ebonite, Thiokol, and other polymers shaving room air

containing sulfide bonds in the main chain compression room air
shaving 77 liqu. N2 + 10% 02
28. Polydimethylsilmethylene drilling 77 liqu. N2 none

-Si(CH 3h- CH r
29. Polyoxymethylenedimethylsiloxane drilling 77 liqu. N2 none

-Si(CH3h-CH2-0-CH2-0-
-CH 2 -Si(CH 3h-0-
30. Polydimethylsilphenylene-dimethyl siloxane drilling 77 liqu. N2 none

-Si(CH3h-C6H4-Si(CH3h-0-
31. Polydimethylsildiphenylene- drilling 77 liqu. N2 none

oxide dimethylsiloxane
-Si(CH 3)2- C6H4-0-C6H4Si(CH3h-0-
32. Polydimethylsiloxane drilling 77 liqu. N2 none

-Si(CH 3h- 0 -
--
33. Polydimethyl-(methylvinyl)- drilling 77 liqu. N2 none

siloxane
-Si(CH3h-0- 99.9%
34. Polydimethyl-(methylethyl)- drilling 77 liqu. N2 none

siloxane
-Si(CH 3h-0- 92%
-Si(C2HS)(CH 3)-0- 8%
35. Polydimethylsiloxane drilling 77 liqu.N 2 none

(crosslinked polymer)
-Si(CH3h-0- 98.5%
-Si(O- )(CH3)-0- 1.5%
134
1 V. Assignment or ;::,pectra

spectra Ref.
K
25 77 -S-CH2 -CH2-S-CH-CHr 34
26 . (?) no R-S 34
27 .room I
R-S 5,34,35
77
77
J
28 77 traces of =Si-CH 2 ROO 36
29 77 traces of =Si-CH 2 ROO 36
30 77 none ROO and traces of=Si-CH 2 36
31 77 none ROOand traces of=Si-CH 2 36
32 77 none none (ionic degradation) 36
33 .77 none none (ionic degradation) 36
135
Table 6.2. Form of Spectrum and Concentration of Free Radicals Formed in Tension
Polymer Sample and treatment Environment
1. Polyethylene bundle of filaments hexane at 243 K

yarn, 300 filaments, 0.15 mm diam. N2
moulded sheet, 3.5 mm thick N2 + O2 , 160 to 294 K
2. Polypropylene yarn, 200 filaments, 0.20 mm N2
diameter
3. Acrylonitrile-butadiene cut sheets 2.8 mg 03/1
100% preoriented at 300 K N2 at 118 to 168 K
4. chl-Polyisoprene vulcanized sheet, 2 mm thick
(natural rubber) prestretched; sulfur-cured; Ar + N2 at 93 K
dicumyl-peroxide cured Ar+ N2 at 93 K
5. Polychloroprene dicumyl-peroxide cured, Ar + N2 at 93 K
100% preoriented at 300 K
6. Polystyrene stack of strips hexane at 243 K
film, 1. 3 mm thick (air? )
7. Styrene copolymers film, 1. 3 mm thick (air?)
8. Polymethylmethacrylate bundle of filaments hexane at 243 K
9. Polyethylene terephthalate bundle of filaments hexane at 243 K
yarn, 7000 filaments, 0.02 diameter N2
(mono filament?) air at room temperature
10. Polycarbonate bulk, drawn to necking air at room temperature
11. Polycaprolactam bundle of fibers hexane at 243 K
bundle of fibers vacuum
bulk material vacuum
yarn, 7000 filaments, 0.02 diameter N2
(monofilament?) air at room temperature
monofilament air at room temperature
12. Poly(hexamethylene adip- preoriented film, 0.5 mm thick air at room temperature
amide) (66 polyamide)
13. Poly(laurolactam) yarn, 6800 filaments, 0.02 diameter N2
(12 polyamide)
14. Polyoxamide bundle of fibers air at room temperature
15. Natural silk bundle of fibers hexane at 243 K
16. Silicone rubber cut sheet 2.8 mg 03/1
cut sheet, 100% prestrain at 300 K N2at194K
17. Poly(p[2 hydroxyethoxy) bundle of fibers vacuum
benzoic acid)
1. P. Ayscough: Electron Spin Resonance in Chemistry: Methuen, London 1967.

2. B. Ranby, J. Rabek: ESR Spectroscopy in Polymer Research. Heidelberg: Springer-Verlag
1977.
3. H. Fischer: Magnetische Eigenschaften freier Radikale: Landolt-Bornstein-Tabellen, Neue
Serie II/I, K.-H. Hellwege, A. M. Hellwege, Eds., Berlin: Springer-Verlag 1965.
136
ESR spectra Observed radical Concentration Ref.

taken at K spins/cm 3
1. 77 peroxy (R-O-O) 5.10 16 45

291 singulet 5 to 50'10 14 37,46
103 peroxy 1.6 to 48.10 15 47
2. 291 none 10 14 37,46
3. room asymmetric triplet up to 10 18 48

118 to 168 asymmetric triplet up to 2.10 16 49
4. asymmetric qui~tet
113 (-CH 3 C=CH-CH2?) 4toll'10 17 49,50
113 peroxy well observable 50
5. 113 peroxy 51
6. 77 none 10 14 45
(room) none 10 11 * 52
7. (room) none 10 13 52
8. 77 none 10 14 45
9. 77 peroxy 8.10 15 45
291 none 10 14 37,46,53
113 peroxy up to 15 . 10 16 53
10. room ill resolved quartet up to 1.6' 10 16 54
11. 77 quintet (-CO-NH-CH-CH2-) 5' 10 17 44,45
298 quintet 1 . 10 17 55
298 ~one 55
291 quintet l' 10 18 28,32,37,46
113 peroxy up to 7' 10 16 53
77 quintet 5' 10 16 56
12. 300 triplet (unresolved quintet?) 57
13. 291 quintet 5 . 10 14 to 5 '10 15 37,46
14. 300 quintet 1 . 10 16 58

15. 77 doublet 7 . 10 17 44,45
16. room multiplet up to 2' 10 16 48
194 multiplet up to 2' 10 16 59
17. room slightly asymmetric triplet well observable 60
* (indirectly estimated)
4. S. E. Bresler, S. N. Zhurkov, E. N. Kazbekov, E. M. Saminskii, E. E. Tomashevskii: Zh.

tekhn. fiz. 29,358 (1959), Soviet Physics - Techn. Physics 4, 321-325 (1959).
5. S. E. Bresler, E. N. Kazbekov, E. M. Saminskii: Polymer Sci. USSR 1,540 (1959).
6. P. Yu. Butyagin, A. A. Berlin, A. E. Kalmanson, L. A. Blumenfeld: Vysokomol. Soyed.
1, 865 (1959).
7. P. Yu. Butyagin, A. M. Dubinskaya, V. A. Radtsig: Uspekhi khimii38, 593 (1969). Russian
Chern. Rev. 38, 290- 305 (1969).
137
8. H. H. Kausch: Rev. Macromol. Chern. 5(2),97 (1970).

9. K. L. DeVries, D. K. Roylance, M. L. Williams: Report from the College of Engng., Univ.
of Utah, Salt Lake City, UTEC DO 68-056, July 1968.
10. V. A. Zakrevskii, E. E. Tomashevskii, V. V. Baptizmanskii: Fizika Tverdogo Tela 9,
1434-1439 (1967); Soviet Physics - Solid State 9, 1118-1122 (1967).
11. V. A. Radtsig, P. Yu. Butyagin: Vysokomol. Soyed. A 9, 2549 (1967); Polymer Sci.
USSR 9, 2883 (1967).
12. V. A. Zakrevskii, V. V. Baptizmanskii, E. E. Tomashevskii: Fizika Tverdogo Tela 10, 1699
(1968), Soviet Physics - Solid State 10, 1341-1345 (1968).
13. J. Sohma, T. Kawashima, S. Shimada, H. Kashiwabara, M. Sakaguchi: Nobel Symposium
22, Almquist & Wicksell, Stockholm, 225-233 (1972).
14. T. Kawashima, S. Shimada, H. Kawashibara,J. Sohma: Polym.J. (Japan) 5, 135-143 (1973).
15. H. Yamakawa, M. Sakaguchi, J. Sohma: Rep. Progr. Polymer Physics, Japan, XVI, 543
(1973). M. Sakaguchi and J. Sohma: Rep. Progr. Polymer Physics, Japan, XVI, 547 (1973).
M. Sakaguchi, H. Yamakawa and J. Sohma: Polymer Letters 12, 193 (1974).
16. P. Yu. Butyagin, I. V. Kolbanev, V. A. Radtsig: Fizika Tverdogo Tela 5, 2257-2260,
Soviet Physics - Solid State 5, 1642-1644 (1964).
17. A. M. Dubinskaya, P. Yu. Butyagin: Vysokomol Soyed. A 10, 240-244 (1968), Polymer
Sci. USSR 10,283-288 (1968).
18. A. M. Dubinskaya, P. Yu. Butyagin: Vysokomol. Soyed. B 9,525 (1967).
19. S. N. Zhurkov, E. E. Tomashevskii, V. A. Zakrevskii: Fizika Tverdogo Tela 3, 2841-2847
(1961), Soviet Physics - Solid State 3, 2074 (1962).
20. J. Tino, M. Capla, F. Szocs: European Polymer J. 6, 397 (1970).
21. A. M. Dubinskaya, P. Yu. Butyagin, R. R. Odintsova, A. A. Berlin: Vysokomol. Soyed.
A 10,410-415 (1968), Polymer Sci. USSR 10,478-485 (1968).
22. V. A. Radtsigand P. Yu. Butyagin: Vysokomol. Soyed.A 7,922 (1965), Polymer Sci. USSR
7, 1018 (1965).
23. P. Yu. Butyagin: Dokl. Akad. Nauk 140, 145 (1961).
24. P. Yu. Butyagin, I. V. Kolbanev, A. M. Dubinskaya, M. Yu. Kisljuk: Vysokomol. Soyed.
A10, 2265-2277 (1968).
25. S. A. Kommissarov, N. K. Baramboim: Vysokomol. Soyed. All, 1050-1058 (1969),
Polymer Sci. USSR 11, 1189-1199 (1969).
26. V. A. Zakrevskii, E. E. Tomashevskii: Vysokomol. Soyed. B 10, 193 (1968).
27. T. Nagamura, M. Takayanagi: J. Polymer Sci., Polymer Phys. Ed. 13, 567 (1975).
28. D. K. Roylance, K. L. De Vries, M. L. Williams: P. II. Internat. Conf. Fracture, Brighton,
April 13-18, 1969.
29. D. K. Roylance: Ph. D. Dissertation, Dept. of Mech. Engng., Univ. of Utah, Aug. 1968.
30. D. K. Backman: Ph. D. Thesis, Univ. of Utah, May 1969.
31. D. K. Backman, K. L. De Vries: J. Polymer Sci. A-I 7, 2125 (1969).
32. M. L. Williams, K. L. De Vries: 14th Sagamore Army Materials Res. Conf. Aug. 23 (1967).
33. K. L. De Vries, D. K. Roylance, M. L. Williams: J. Polymer Sci. A-2, 10,599 (1972).
34. V. A. Zakrevskii, E. E. Tomashevskii: Polymer Sci. USSR 8, 1424 (1966).
35. S. N. Zhurkov, V. A. Zakrevskii, E. E. Tomashevskii: Fizika Tverdogo Tela 6, 1912-1914
(1964), Soviet Physics - Solid State 6,1508-1510 (1964).
36. K. A. Akhmed-Zade, V. V. Baptizmanskii, V. A. Zakrevskii, S. Misrovand E. E. Tomashevskii:
Vysokomol. Soyed.A 14,1360-1364 (1972), Polymer Sci. USSR 14,1524-1530 (1972).
37. J. Becht: Dissertation, Technische Hochschule, Darmstadt (1971).
38. H. H. Kausch, J. Becht: Kolloid-Z. u. Z. Poly mere, 250, 1048 (1972).
39. T. Pazonyi, F. Tudos, M. Dimitrov: Angew. Makrom. Chemie 10,75-82 (1970).
40. H. Fischer: Polymer Letters 2, 529 (1964).
41. J. B. Park: Ph. D. Thesis, Univ. of Utah, June 1972.
42. C. T. Graves, M. G. Ormerod: Polymer 4, 81 (1963).
43. M. Kashiwagi: J. Polymer Sci. A 1,189 (1963).
44. S. N. Zhurkov, A. Ya. Savostin, E. E. Tomashevskii: Dokl. Akad. Nauk. SSR, 159, 303
(1964), Soviet Physics - Doklady 9,986 (1964).
138
45. S. N. Zhurkov, E. E. Tomashevskii: Proc. Conf. Physical Basis of Yield and Fracture,
Oxford 1966, p. 200.
46. J. Becht, H. Fischer: Kolloid-Z. u. Z. Poly mere 240, 766 (1970).
47.1. A. Davis, C. A. Pampillo, T. C. Chiang: Bull. Amer. Phys. Soc.IB, 433 (1973), J.
Polymer Sci., Polymer Phys. Ed. 11, 841-854 (1973).
48. K. 1. De Vries, E. R. Simonson, M. 1. Williams: J. Macromol. Sci. - Phys. B-3, 671
(1970).
49. R. T. Brown, K. 1. De Vries, M. 1. Williams: Polymer Preprints 11/2, 428 (1970).
50. R. Natarajan, P. E. Reed: J. Polymer Sci. A-2, 10,585 (1972).
51. E. H. Andrews, P. E. Reed: In: Deformation and Fracture of High Polymers: H. H.
Kausch, 1. A. Hassell and R. I. Jaffee, Eds., Plenum Press, New York 1973, p. 259.
52. 1. E. Nielsen, D. J. Dahm, P. A. Berger, V. S. Murty, J. 1. Kardos: J. Polymer Sci.,
Polymer Phys. Ed. 12, 1239 (1974).
53. T. C. Chiang, J. P. Sibilia: J. Polymer Sci., Polymer Phys. Ed. 10, 2249 (1972).
54. B. Yu. Zaks, M. 1. Lebedinskaya and V. N. Chalidze: Vysokomol. Soyed. A 12, 2669
(1970), Polymer Sci. USSR 12, 3025 (1971).
55. D. Campbell, A. Peterlin: Polymer Letters 6,481 (1968).
56. B. Crist, A. Peterlin: Makromol. Chern. 171, 211-227 (1973).
57. G. S. P. Verma, A. Peterlin: Polymer Preprints 10, 1051 (1969).
58. P. Matthies, J. Schlag, E. Schwartz: Angew. Chern. 77, 323 (1965).
59. R. T. Brown, K. 1. De Vries, M. 1. Williams: Polymer Letters 10, 327 (1972).
60. T. Nagarnura, M. Takayanagi: J. Polymer Sci., Polymer Phys. Ed. 12, 2019-2034 (1974).
61. J. Ham, M. K. Davis, J. H. Song: J. Polymer Sci., Polymer Phys. Ed. 11, 217 (1973).
62. M. Sakaguchi, J. Sohma: J. Polymer Sci., Polymer Phys. Ed. 13, 1233-1245 (1975).
63. M. Sakaguchi, S. Kodama, O. Edlund, J. Sohma: J. Polymer ScL, Polymer Letters Ed.
12,609 (1974).
64. J. Sohma, M. Sakaguchi, in: Adv. Polymer ScL, Vol. 20, Berlin, Heidelberg: Springer-
Verlag 1976.
65. V. A. Radtzig: Vysokomol. SoyedA 17,154-162 (1975). Polymer Sci. USSR 17,
179-190 (1975).
66. J. Tino, 1. Placek, F. Szocs: European Polymer J. 11,609-611 (1975).
67. D. Campbell: J. Polymer Sci, Part D, Macromol. Rev. 4,91-181 (1971).
139
Chapter 7
Phenomenology of Free Radical Formation

and of Relevant Radical Reactions ( Dependence
on Strain, Time, and Sample Treatment)
I. Radical Formation in Thermoplastics . . . . 141

A. Constant Rate and Stepwise Loading of Fibers 141
B. Effect of Strain Rate on Radical Production 150
C. Effect of Temperature 152
1. Apparent Energy of Bond Scission 152
2. Rate of Bond Scission . 154
3. Concentration at Break 156
D. Effect of Sample Treatment 159
II. Free Radicals in Stressed Rubbers 162
A. Preorientation, Ductility, and Chain Scission 162
B. Cros9-Link Density, Impurities, Fillers 165
Ill. Mechanically Relevant Radical Reactions 167
A. Transfer Reactions . 168
B. Recombination and Decay 169
C. Anomalous Decay. . 170
D. Radical Trapping Sites. . 170
I. Radical Formation in Thermoplastics
A. Constant Rate and Stepwise Loading of Fibers
Historically, the first ESR experiments with mechanically produced free radicals
have been made in the Ioffe-Institute in Leningrad in 1959 [1] with ground or milled
polymers, the samples being investigated after completion of the fracture process.
To elucidate the effect of structural or environmental parameters on the kinetics of
stress-induced free radical formation it is necessary to study - by ESR - highly
stressed chains during the loading process. As outlined in Chapter 5 a notable energy
elastic straining of a chain can only be achieved if the chain cannot relieve stresses
internally by a change of conformation or externally by slippage against the field
exerting the uniaxial forces. Vice versa the mechanical scission of a chain must be
taken as an indication that at the time of chain rupture axial stresses 1/1 equal to the
chain strength 1/1 c had not only been reached but also maintained during the average
lifetime period T c of the chain segment.
141
7. Phenomenology of Free Radical Formation
In view of the fact that the slippage of chains, microfibrils, or fibrils reduces or
prevents the mechanical scission of chains highly oriented thermoplastic fibers with
little potential for plastic deformation are the most suitable objects for a study of
chain scission kinetics. ESR investigations using fibers of polyamide 6, 66, and 12,
PE, PP, PETP and of other materials have been carried out in a limited number of
laboratories, at first in the USSR, later in the USA, Germany, Great Britain, and
Japan (see Table 6.2). Practically all researchers worked with highly oriented mono-
filaments, stacks of strips, or mostly with commercial yarns made up of a few hun-
dred filaments each having a diameter of about 20 Ilm. As discussed in Chapter 2 the
filaments consist of fibrils and these in turn of microfibrils with a diameter of 20
to 40 nm. The experimental set-up allowing uniaxial straining of these samples under
controlled atmosphere and directly within the ESR cavity has been described and
referenced in Chapter 6, Section IV B. It has also been seen that the primary radicals
formed in tensile straining can - with one exception - be assigned to the breakage
of main chains. In the following discussion, concentrations of secondary - rather
than of primary - radicals are usually analyzed. It can be understood that the con-
version factor from the number of primary radicals to the number of secondary radi-
cals is constant (and, if not stated otherwise, close to unity).
If a semi crystalline microfibril is subjected to stresses, the resulting deformation
will be non-homogeneous at the molecular level. The bulk of the deformation will
be born by the amorphous regions. As discussed in Chapter 5 the largest stresses are
transferred upon extended chain segments which share the strain imparted onto an
amorphous region. The stress induced scission of chains must, therefore, be ex-
pected to occur within the amorphous regions.
Becht and Fischer [2] have demonstrated directly that free radicals are formed
in the amorphous regions. These authors observed that polycaprolactam samples
swollen in methacrylic acid did not show in subsequent stressing the usual ESR
spectrum of the polyamide radical but the polymerization radical of the methacrylic
acid. Based on the logical assumption that only amorphous regions had been swollen
it was proved, therefore, that free radicals had formed in the amorphous regions
only. Verma et al. [3] reached at the same repeatedly confirmed conclusion studying
irradiation-produced radicals in semi crystalline polymers. Those radicals had been
created under 'Y-irradiation within the whole volume of a PA 66 film i.e. as well with-
in amorphous as within crystalline regions. At room temperature Verma observed -
depending on sample orientation within the ESR cavity - a three, four, or six line
spectrum. He assigned the pronounced anisotropy of the spectrum to the fact that
the majority of the remaining radicals is located in the well oriented crystalline blocks.
If the free radicals were introduced into the same material by stretching, no conspi-
cuous anisotropy of the ESR spectrum could be observed. In this case the radicals
were obviously located in sites showing rather small local chain orientation. In
terms of fiber morphology the chain scission events must have taken place, there-
fore, in the amorphous regions between the crystal lamellae within the microfibrils
(Fig. 2.11, 2.17b, 2.18b), at microfibril surfaces, or in interfibrillar regions (Fig. 2.11).
The question of the location of chain scission points will be touched upon repeated-
ly within this chapter.
At first the phenomenology of free radical production in preoriented yarns will
be discussed taking 6 and 66 polyamide as examples. In constant load rate tests a
142
very steep increase in concentration of fairly stable secondary radicals is observed

in a strain regime of between 8% and rupture strain (16 to 25%) which corresponds
to a stress regime of between 500 and 900 MN/m 2 (Fig. 7.1). Load rate tests are a
suitable means to demonstrate the effect of mechanical action on radical formation
and accumulation, but they are not sufficient to discriminate between the simulta-
neous effects of stress, strain, and time on the rate of radical formation. From con-
stant load tests it was learned that, under constant load, the rate of radical forma-
tion decreases quite rapidly whereas the amount of radicals formed in virgin samples
reaches a constant equilibrium value (Fig. 7.2). If a sample is strained repeatedly to
a strain level tJ then free radicals will be formed, i.e. chains will break, only during
the first straining cycle. If, however, in subsequent straining the level tJ is exceeded,
more chains will be seen to break. In a step-strain test the radical population increases
stepwise (Figs. 7.3 and 7.4).
150 1000
'"S 100 "'E

x Z
.~ 500 :<:
l
50
l 0
0 40 80 120 160 0
16
'"c;
~
00 40 80 120 160
150
-
[id
"'
S
x
~
100
I
~
c::
.~ 50
Fig. 7.1. Typical stress, strain, and free radical

00 40 formation curves in constant load-rate strain-
time,s ing of 6 polyamide fibers [41.
In the first paper on this subject [7] Kausch and Becht gave a detailed discussion
of the notable fact that within each strain step the radical concentration increases
concurrently with a decrease of the macroscopic tensile stress. The conclusion was
drawn that the local molecular stresses are unevenly distributed. They are not simply
a constant multiple of the macroscopic stresses. The chain segments breaking to-
wards the end of a strain interval quite obviously do not participate in the macro-
143
4OOrr=========----,
.=! 1000
200 ;z
"16'
.£
0 ~----~----~----~--~ro 0
0 20 40 60
15.0
'"c· 1.5
~
0.00 20
120
(il l
...-9 ro
~
~
c 40
.~
Fig. 7.2. Load, strain, and free radical pro-

duction in a constant-load test of 6 poly-
00 20 40 60 amide fibers [41.
time, S
Fig. 7.3. ESR spectra of secondary radicals produced

after stepwise loading of 6 polyamide fibers. The num-
bers correspond to stress levels of 0,690, 760, 794,
828, 876, and 897 MN/m2 respectively [5 J.
144
600
T: 18° C
200
{R}
2 "",.c<> Fig. 7.4. Concentration of free
t radicals formed in stepwise
>"'£"
113.5114.2114.9115.5116.1 116.8 ~7."'%1 t
O+-"'--:....-"":-....-..r-"'--::"-"":"...---i-""""':.............;..J--
straining of 6 polyamide fibers
and measured uniaxial stress as
o 20 "'0 60 80 100 120 min a function of strain and time [61.
scopic stress relaxation (a decrease of the axial chain stress to 0.9 t/lc increases the
lifetime by two orders of magnitude and virtually stops the chain scission process).
It was furthermore concluded that the microfibrils do not unload through slippage
and that the lateral rigidity of the crystal blocks within the sandwich-structured
microfibrils must be large enough so as to permit the build-up of large stress concen-
trations. If the lateral rigidity of the crystal blocks were insufficient the fibril would
break up into submicrofibrils. This situation is schematically shown by Figure 7.5. The
i 0(1) 0=0
E·
0 0 = ~ 0(1)
Ej
I.
Fig. 7.5. Model of the micro-deformation and resulting local stresses within different amorphous
(u a) and crystalline (u c ) regions of semicrystalline fibers; case I. large, case II. small lateral
rigidity of crystalline phases; Ei: Young's moduli of amorphous subregions, supposedly depending
on the concentrations of highly extended tie molecules which are schematically indicated by
the number of heavy lines between crystalline regions.
145
central diagram represents the alledged sub microfibrillar structure within an un-
stressed microfibril. If the crystal blocks break up while being strained the tensile stresses
in every part of the structure will equilibrate and be equal to the external stress
(right hand side). The external stresses are, as one knows from Chapters 1 and 5 al-
ways much smaller than the chain strength. If, however, the layered structure is also
preserved at large strains stresses proportional to local strain and tensile modulus Ej
of the amorphous subregions can be exerted (left hand side of Fig. 7.5). This means
that within a subregion with large Ej correspondingly large - although locally well
confined - stresses €aEj - can be reached. For this reason the left hand side model
is proposed to describe the mechanical response of an axially strained microfibril.
The observed fact that chains break even after (20 minutes of) stress relaxation
not only demands the integrity of the crystal blocks but also an intimate and per-
sistent lateral cohesion between microfibrils within a fibril and between fibrils within
a filament. In the same way as studied in detail in Chapter 5 for single chains the
shear displacement of the ends of microfibrils against the interfibrillar cohesion per-
mits the transfer onto the microfibrils of those forces which accumulate within the
stress-transfer lengths to the axial stress a. This stress relaxes at constant filament
elongation. The continued rupture of chains indicates that the axial strains of the
microfibrils are maintained during such stress relaxation. Those strains, however,
can only be maintained if large scale slip of microfibrils or fibrils does not occur.
In the described experiments with PA 6 fibers the cause of stress relaxation must
obviously be sought within the microfibrils. Following the model shown on the left
hand side of Figure 7.5 decrease of aCt) at constant microfibril elongation can be
caused.
- either by the gradual and fairly general decrease of the relaxation moduli Ej of
practically all amorphous regions within the microfibril
or by the preferential and large decrease of the Ej of just a few amorphous regions,
accompanied by their large deformation and by the retraction of others.
The previously discussed intimate lateral cohesion would favor the more homo-
geneous decrease of all relaxation moduli. The same conclusion was drawn by Geza-
low et al. [8] from the uniformity of stress induced long period changes.
The following statements will serve, therefore, as the basis for a mathematical
analysis of the ( static) chain scission experiments such as step-strain or small loading
rate tests at sufficiently low temperatures:
stress relaxation or creep is not caused by slippage of fibrils or microfibrils;
chain scission occurs in tie-segments, i.e. in those segments of a chain molecule
which connect two different crystal blocks traversing in a fairly extended con-
formation the amorphous region in-between (e.g. Fig. 5.2); most frequently tie-
segments interconnect adjacent crystal blocks within a microfibril, but also more
distant or interfibrillar ties have to be considered;
eventually breaking tie-segments are solidly held, most probably by the crystal
blocks; although the clamping extends over 2 to 5 nm the tie-segments are treated
as if they had a well defined end-to-end distance Lo and contour length L;
chain scission occurs if the axial chain stress 1/1 reaches chain strength 1/1 c;
the large value of molecular stress concentration (ratio of 1/1 to a of the order of
50) follows from the fact that individual (highly rigid and nearly extended) chains
146
are subjected to the comparatively large strains of the intra- or interfibrillar amor-
phous regions they span.
At this point only the results of the mathematical analysis with respect to the
kinetics of the formation of free radicals are presented. A discussion of the possible
role of chain scission in fracture will be resumed in Chapter 8.
The analysis concerns the general case of fairly extended tie-segments of contour
length ~, end-to-end distance Lo, and elastic modulus E k . Each segment spans an
arbitrary number nai of amorphous regions (width La) and nei = na - 1 of crystalline
regions (width Lc). As discussed in Chapter 5 a segment i will experience ultimate
stresses 1/Ii only after it has assumed the most extended conformation accessible to
it. The elastic stress then is:
(7.1 )
At a given average axial microfibril strain E the stresses 1/Ii are largest for tie-
segments interconnecting adjacent crystal blocks (Lo = La). The third term in Eq.
(7.1) then disappears. If the tie-segments extend over several sandwich layers and
crystal blocks of the same or different microfibrils the first term in Eq. (7.1) be-
comes small. In that case the average tensile strain E determines the stress level1/li'
It is important to point out again that in this analysis no loading of interfibrillar
tie-segments through shear strains (slippage between microfibrils) will be considered.
At a degree of crystallinity of 50% E is by a factor of (2-E/Ec) smaller than Ea, where
E and Ec deSignate the relaxation modulus of the microfibril as a whole and that of
the crystalline regions.
The mathematical analysis of the free radical production curves aimes at a deter-
mination of the distribution No(~/Lo) and of the kinetic parameters Uo , wo, and
{3. It can be carried out taking into consideration only tie-segments between adjacent
crystal blocks (Lo = La). If only long interfibrillar tie-segments were present the
resulting distribution N~ (~/Lo) would be narrower and approximately obtainable
through the transformation NoI (Li)
- = No ([I+E
--- a ]L-i ) . The following analysis
La [1+E]Lo
also assumes a homogeneous distribution of amorphous strains within a microfibril.
Consideration of a non-homogeneous distribution of strains Eai would again result
in narrowing No(LdLo). From Eqs. (5.57) and (7.1) one obtains straight forwardly
the expectation value of the number of broken chains or formed free radicals:
(7.2)
where Wb = Wo exp (-Uo/RT).

Through a numerical iteration procedure, a distribution of LdLo can be deter-
mined [7] in such a way that Eq. (7.2) describes the experimental points (e.g. those
radical concentrations shown in Fig. 7.4). The length distribution of the intrafibrillar
tie-segments obtained in that case is shown in Figure 7.6, it ranges from L/Lo = 1.06
147
/--1
// i
I i
U I
Ii;
Q.
.0:
V! 1 effect of double kink
(5nm PA chain)
y ! '/.~
Z
2
!
I
IT (1.1 i m(98.~'/.)
, I
.A
V(O.3"i IL/~o=1.18 I =+L/L O
~ ~ m ~ ~u~
Fig. 7.6. Distribution of relative chain lengths of tie molecules in the amorphous regions of
6 polyamide fibers as obtained from numerical evaluation of step-strain ESR data [4, 71.
I chains broken until fracture, II chains fully extended and contributing with chain modulus
Ek to sample modulus E, III chains in non-extended conformation.
to 1.18. If the same experimental radical concentration points were to derive from
the rupture of fairly long interfibrillar tie-segments then their relative lengths must
have a distribution of L/Lo between 0.97 (!) and 1.05. Since a given distribution of
segment lengths is essentially reflected by the step heights of radical eqUilibrium
concentrations a histogram of these step heights serves to characterize the apparent
segment length distribution without introducing any assumptions on the nature of
the tie-segments. In Figure 7.7 histograms from step strain tests carried out at dif-
ferent temperatures are reproduced.
The association of the observed free radical distributions with apparent chain
length distributions requires the smallest number of additional assumptions. If it
were to be assumed that the microfibrils contained a monodisperse distribution of
120
a b c d
M
E
u
.;;- 80
-52
",-
c
'a
OIl
<I 40
0
0 12 )60 4 8 12 160 4 8 12 4 8 12 16
sample strain, %
Fig. 7.7. Distributions of relative chain lengths shown as histograms of free radical concentrations
obtained in step-straining of 6 polyamide fibers at (a) -25°C, (b) room temperature, (c) + 50°C,
and (d) +100 °c [4,51.
148
1. Radical Formation in Thermoplastics
relative segment lengths (e.g. LILo = 1.0) it were also necessary to assume that
the amorphous strains are unevenly distributed throughout the fiber
the molecular extensibility is comparable to the amorphous strains, i.e. reaches 25
to 35%
there is a rapid strain hardening of an amorphous region after breakage of the
tie molecules (otherwise collapse of that region would lead to complete stress
relaxation in the whole microfibril).
Particularly the latter two conditions seemed to be unrealistic. So the early
proponents [3-7] were led to their interpretation that a fiber contains tie chains of
different conformations ("tautness").
A third interpretation of step-strain data was advanced by Crist and Peterlin [9].
They suggested that there is an uneven distribution of strains due to differences in
the lengths of the many thousand filaments strained simultaneously in any of the
above experiments. The uneven-length effect undoubtedly broadens the observed
distributions of relative chain lengths. But the premature breakages of individual
filaments, and the formation of their fracture surfaces, cannot account for the num-
ber of free radicals formed. To further disprove this contention Hassell and DeVries
examined the free radicals formed in straining of highly oriented bulk nylon 66
banding material [10] . They obtained quite similar and even slightly wider histograms
in straining bulk strip material than bundles of nylon 66 fibers. A scanning electron
micrograph of the strip material fracture surface reveals that in banding material -
as postulated for yarns - defects are introduced throughout the stressed sample
volume (Figs. 7.8 and 7.9). The final fracture surface propagates along the path of
least resistance which includes the previously developed defect zones. It is only at
a level of deformation close to the rupture strain that a difference between the strain-
ing of a single fiber and of a bundle thereof becomes noticeable. A statistical explana-
tion of this fact has been given in Chapter 3.
Fig. 7.S. Fracture surface of dry Fig. 7.9. Fracture surface of 66 polyamide banding
66 polyamide banding material[ 11 0 1. material, before fracture stored to equilibrium at 65%
relative humidity [10 1.
149
Convincing evidence against the assumption that segment length distributions

are macroscopic artefacts was derived from step-temperature tests carried out by
Johnsen and Klinkenberg [11] which will be discussed in Section C.
B. Effect of Strain Rate on Radical Production
Local molecular stresses can partly be relieved by slip or uncoiling of molecules. In

thermoplastics the relaxation times governing those viscoelastic deformations at room
temperature are of the order of milliseconds to minutes, i.e. smaller or comparable
to the duration of the mechanical excitation. Rapid loading, then, can bring more
nonextended chain segments to high stresses and eventually to chain scission than
slow loading during the same strain interval. But their number should be small. On
the other hand, the higher rigidity of a rapidly loaded viscoelastic body leads to a
decrease of the strain at fracture and, therefore, also to decrease of the largest strain,
Ea , imposed on the amorphous regions.
The effect of strain rate on chain scission of rigidly clamped segments follows
from Eq. (7.2). The higher the loading rate the smaller the time t necessary to reach
a certain strain, the shorter also the time interval a stressed chain spends at the stress
level aCt) and the smaller the probability for chain scission at that level [7]. In
Figure 7.10 the decay curves of polyamide 6 segments of monodisperse contour
lengths L = 1.09 Lo have been plotted. At 20% amorphous strain Ea a segment of
length L will assumedly have a fully extended conformation and experience a chain
strain of 10%. At this strain level chain scission becomes noticeable at small strain
rates (0.1 . 10- 6 S-I). At larger strain rates an equivalent decay is observed later.
For the given system an increase of strain rate by a factor of 10 will shift the decay
curve to higher strain values by .::lEa = 0.9%.
The effect of a change of strain rate in the course of a tensile experiment can
be predicted from Figure 7.10. An instantaneous increase in strain rate is equivalent
"~o
----{ r
100
·iii
III €G O 0.0006
·u
0.006 -
-
III
c
..
!!
0.06 --
- - -.......
.~
r
E 0.6 'Io/min ..
c 50
~u A ,
"0
~
f0. 0
o 5 10 15 20
)))) 25 %
uniaxial strain In amorphous regions, Eo
Fig. 7.10. Effect of rate of strain on the scission of polyamide 6 segments of contour length
L = 1.09 Lo (after 12.7).
150
to changing from a certain level of depletion (e.g. 40% at point A) to a level where
the depletion at that rate should have been much lower (6% at B). In other words a
fraction of 34% of those chains which are supposed to scission beyond B are already
gone. Consequently the rate of chain scission following an increase in strain rate
should be reduced. This has been quantitatively verified by Klinkenberg [13].
Only few experiments are known where the concentration of free radicals as a
function of strain rate has been investigated [12-16]. In Figure 7.11 the total num-
ber of spins at fracture of a 6 polyamide fiber sample is shown to increase in the
range of clamp displacement rates between 1 and 10 . 10- 3 cm/s. Above
20· 10- 3 cm/s (which corresponds to a strain rate of 1.43· 10- 3 S-I) a drop in
free radical production is indicated [14]. The total rate effect seems to be not very
large, however. For the present experiment, it did not exceed 50%. Similar values
have been reported [16] from an investigation of the cold drawing of polycarbonate
[15]. Here the total concentration of free radicals obtained increased by 140% with
strain rates varying from 0.017 S-1 to 1.7 S-1 (Fig. 7.12). The effect of strain rate
on the formation of free radicals in polychloroprene will be discussed in Section II A.
strain rate, 10- 3 em Is
50
20
<I)
I
)(
'"..,E
~
10
..
......
..
c:
'a
Fig. 7.11. The effect of strain

0
0.02 0.04 0.1 0.2 0.4 rate on the radical concen-
tration at break of 6 polyamide
strain rate, in./min. fibers [141.
o~~----~~--~~--~~ ____ ~~ ____ _______

~
strain rate Eo, sol

Fig. 7.12. Effect of strain rate on the concentration of free radicals after necking of polycar-
bonate 115, 161.
151
In all of these cases the variation of free radical concentration seems to reflect
more the changes of material response than the kinetics of decay of stressed chains.
The effect of strain rate on radical production in comminution (grinding, milling)
would be very difficult to determine and no results have been reported.
c. Effect of Temperature
1. Apparent Energy of Bond Scission
It is presumed that the stress-induced scission of chain molecules is achieved through

the cooperative action of mechanical forces (lowering the bond's binding potential
barrier) and of statistically fluctuating thermal vibrations supplying the remaining
increment of energy which is necessary to split the loaded bond. It is also believed
that Eq. (5.57) provides an adequate description of how mechanical and thermal
energy affect the rate kc of the thermo-mechanical chain scission process. If this
presumption is correct, the lack of thermal vibrational energy should increase the
stability of a stressed bond. Vice versa an increase in thermal energy should make
previously stable bonds reach the critical level of excitation and should break them.
It is of interest to quantitatively analyze this aspect of interaction of thermal and
mechanical energy terms in the scission kinetics of 6-polyamide chains.
The largest energy available to one RCONHCH 2 -CH 2 R' bond within a time
period of l/kb is (Eq. 5.57 for ne = 1):
(7.3)
Using as before a value of ""0 equal to 10 12 S-l, a value of ~ of 5.53· 10- 6

m 3 /mol, and a rate kb = l/Tb = 1/48 S-l the thermal term in Eq. (7.3) has, at a
temperature of 20°C, a value of 77 .2 kl /mol. The stress term ~ 1/1 must reach, there-
fore, a value of 110.8 kl/mol before U(T) is larger than the Uo of PA 6 and bond
rupture at room temperature can occur. At 200 K the thermal term has decreased
to 52.7 kl/mol. In order to provide a total energy of 188 kl/mol, the mechanical
term must now reach 135.3 before bond breakage will occur within 48 s. For a
chain segment containing ne weak bonds the rate of scission, 1/T e' employed in the
above calculations will be I/Te = nc/Tb.
If one assumes that neither ~ nor Uo depend on temperature, one has to conclude
that at 200 K it is possible to raise the molecular stress to a level which is 17% higher
than the largest molecular stress attainable at room temperature. In other words, a
polyamide fiber highly strained at 200 K will quite stable contain many bonds with
~ 1/1 values between 110 and 135 kl /mol. Those bonds will break, of course, if the
temperature is raised to room temperature provided that no other means of stress
relaxation (slip, uncoiling) become available to the strained chain segment.
The search for chain scission events during an upward temperature scan can be
called a critical experiment for the validity of both the kinetic equation (Equation
5.57) and the morpholOgical model (Fig. 7.5). These investigations have been carried
out in the Deutsches Kunststoff-Institut in Darmstadt. Following earlier experiments
152
by Becht [5] Johnsen and Klinkenberg [11, 13] step-strained 6 polyamide fibers at
206,227,248,269 and 290 K. These tests showed the same stepwise increase of
radical concentration and a relaxing stress as reported earlier by them [6, 7, 11, 17]
and other authors [3, 4, 10, 14] and as described in Section A. The result ofa step-
temperature experiment is shown in Figure 7.13. Whereas the stress relaxation is
much more pronounced here than it is in a step-strain test, the radical accumulation
£=14.2%
2.5 750
2.0 700
..... (T
........,.,
0
>< N
1.5 650 E
,.,
'........•
~
......
E z
0 ~
...... 41
<I)
c:: 1.0 600
·50
<I)
0.5 550
0
0 20 40 60 80 100 120
time, min.
Fig. 7.13. Concentration of free radicals and uniaxial stress in step-temperature test as a function
of temperature and time for 6 polyamide fibers [111.
appears to be very similar in both experiments. The authors [11] established that
the final radical concentration [R (T, e)] practically did not depend on the way T
and e were attained. Their experiment has proven the contention made earlier about
the kinetics of thermo mechanical chain scission. It has also proven that the·highly
loaded chains do show in the region T < 260 K essentially no slip or anelastic defor-
mation during the time they are held at the high load, despite the considerable
amount of overall stress relaxation. A slight loss of stressed chains becomes apparent
at T>270 K.
The equivalence of thermal and mechanical action on a given network of chains
is further illustrated by calculating the relative length L/Lo of those intrafibrillar
chains which are just about to break. One considers that condition to be fulfilled
if the 1/J (L), calculated from Eq. (7.1) for La = Lo and introduced into Eq. (7.3),
yields Uo:
(LoL) -1_ +(U

(l - xE/Ec)e/(l -x) + 1
o -RT£nwo!kc)/PE k •
(7.4)
153
E and Ec are respectively the (temperature-dependent) Young's moduli of the

sample as a whole and of the crystal lamellae in chain axis direction, x is the degree
of sample crystallinity.
Equation (7.4) relates sample strain e, temperature T, and relative length L/Lo
of critically loaded chain segments, but subject to the condition that no chain slip
occurs simultaneously. Under this condition the following equivalence between in-
cremental changes of sample strain and temperature is obtained from Eq. (7.4):
.::le=- (I -x).::lT [LRQnwo/kc -{(I + Uo -RTQnwo/kc)~ -I}.

(1 - xE/Ec) Lo (3 Ek (3 Ek Lo
. d(E/Ec)/dT ]. (7.5)
1 - xE/E c
The two terms in the square brackets refer respectively to the change of chain
strength and of strain distribution with temperature. Using the kinetic parameters
employed in evaluating Eq. (7.3) for PA 6 chains, values of Ec = 25 GN/m 2 and of
Ek = 200 GN/m 2 as previously, and an estimated temperature coefficient d(E/Ec)dT
of -0.8· 10- 3 K- 1 one obtains at a temperature of -20°C the following quantita-
tive evaluation of Eq. (7.5):
.::le = - [0.12.10- 3 L/Lo + 0.48.10- 3 (1.1 L/Lo -I)].::l T. (7.6)
This means that for a critically loaded PA 6 chain segment with L/Lo = 1.1 a
temperature increase of 10 K can be compensated by a decrease of strain of only
0.23%. The result of this calculation agrees fairly well with the average value of
0.3%/10 K derivable from the experiments of Johnsen and Klinkenberg [11]. For
preoriented PA 6 fibers they observed that at -60°C a sample strain of 14.6% led
to the same number of (10 17 ) free radicals as a strain of 12.2% at +20 °C.
The loading of possibly present long interfibrillar tie-segments is essentially
determined by 1 + e rather than 1 + ea. Accordingly in Eq. (7 .5) the first term in
square brackets and a front factor equal to .::lT must be retained. For PA 6 this leads to a
.::le equal to -0.24· 10- 3 .::IT L/L o , i.e. to 0.26%/10 K at L/Lo = 1.1. This compari-
son does not permit, therefore, to rule out the existence of interfibrillar tie-segments.
2. Rate of Bond Scission
A rate of bond scission can be defined in terms of time as d [R]/ dt or in terms of

strain as d [R]/ de. The relation between the two rate expressions is defined by the
identity
(d[R])
dt -
Eb E, T
= e(d[R])
de €1,
.
€, T
(7.7)
which has to be taken at constant € and T at corresponding sample strains e1-

154
In the absence of plastic flow (chain scission at strains of 10 to 30%) the rate
can be derived numerically from Eq. (7.2); since the rate will depend on the shape
of the tie-segment length distribution, No (Li)' no general analytical function can
be given for d[R]/de. The effect, however, of a change of € on d[R]/de is the same
as that illustrated by Figure 7.10. The effect of the strain e 1 at which the rate ofbond
scission is determined, depends on the shape of the segment length distribution, i.e.
on the number AN(L/Lo ) of chains having a relative length L/L o suitable to be
broken by an increment Ae of sample strain:
( d[R]) = 2A(L/Lo) (~) . (7.8)

de EJ,
T Ae dL/Lo EJ,
T
In Figures 7.6 and 7.7 typical segment length distributions are given.
A change of temperature will affect the rate of radical production in three ways.
Firstly the number N of highly extended chains will decrease if thermally activated
slip becomes possible. Secondly the activation energy U (T) decreases with increasing
temperature. Thirdly the sample modulus E and the crystalline modulus Ec are af-
fected by temperature.
If in a region of sufficiently low temperatures chain slip can be excluded i.e. if
~/ is constant with T the temperature affects solely the width A (L/Lo) of the
dL Lo
chain segment population which is critically loaded by a strain increase:
A(L/Lo) = (1 - xE/Ec)/(1 - x)
(7.9)
Ae [1 + (Uo - RTQnwo/kc)/~Ek] .
Employing the same values as used in the evaluation of Eq. (7.3) one arrives at
A(L/Lo) = 1.7 .1e at 20°C and at .1(L/Lo) = 1.6.1e at -60°C. This means for
example that a given distribution of segments with relative lengths L/Lo between
1.1 and 1.3 break at room temperature within the strain interval from 12 to 13.5%.
At a temperature of -60°C sample strains of between 13.5 and 15.1% are required.
A completely different chain loading mechanism prevails during the plastic de-
formation of polymers at strains of between 30 and several hundred per cent. In that
case a chain will rupture as a consequence of the frictional forces encountered by
the cham itself or by other morphological units in the dynamic shearing (cf. Chap-
ter 5 II E). The obtainable axial chain stresses are proportional to the molecular or
fibrillar friction coefficients and to the strain rate E. The number of critically loaded
chains will reflect, therefore, the strong increase of the friction coefficient with de-
creasing temperature. Davis et al. [19] strained sheets of high molecular weight poly-
ethylene in air and observed the formation of peroxy radicals. At a level of true strain
£n £/£0 of e.g. 1.1 - which corresponds to a conventional strain of 200% - the con-
centration of the peroxy radicals increased from 5 . 10 14 cm- 3 at 294 K to
10 16 cm- 3 at 160 K. The rate of radical accumulation, d[R]/d £n(£/£o), revealed
two transition regions, one between 180 and 200 K and another beginning at 250 K.
155
The low temperature transition was assigned to the transition of amorphous PE to the
glassy state, the other to slippage of chains within crystal lamellae [19].
3. Concentration at Break
The increase of temperature has a fourfold effect on the number of free radicals ob-
served at the point of macroscopic fracture. Firstly as seen previously the bond
strength is decreased and thus chain scission at a given molecular stress is facilitated.
Secondly the decrease in intermolecular attraction and the increase in molecular
mobility lead to more rapid relaxation of molecular stresses. For the same reason
- thirdly - the density of stored elastic energy at a given strain level is decreased
which will in turn affect crack stability and propagation. Fourthly, the increased
reactivity of the free radicals may lead to an increased discrepancy between the con-
centrations of radicals formed and of radicals observed at the point of failure.
Whenever possible the fourth point has been taken into consideration and radical
concentrations have been corrected for losses due to recombination or other decay
reactions of the (secondary) radicals under consideration. In the case of 6 polyamide
fibers [5, 11, 17, 18] the radical decay is a second order recombination reaction with
the rate constant depending on type, location, and mobility of the radicals (see Sec-
tion III below).
In Figure 7.14 the concentration of free radicals at fracture of 6 polyamide
fibers is shown as observed and also corrected for radical decay during and after sample
fracture [18]. The indicated decrease of the radical concentration at break towards
10
8 o experimental
. 6 corrected
'"E 4
~
c
.~ Fig. 7.14. Concentration of free ra-
dicals at break of 6 polyamide fibers
as a function of breaking temperature
[4,51,0 observed concentrations,
oL---~--~--~--~--~----~
-100 -50 50 100 150 I; concentration corrected for decay
temperatu re T. °c of free radicals.
lower temperatures was well confirmed by Johnsen and Klinkenberg [11] who ex-
tended their measurements to -67 °e. At that temperature they obtained a concen-
tration which was by a factor of 8 smaller than that at room temperature.
The first three of the initially mentioned temperature effects can be better un-
derstood if one makes use of the segment length distribution model. If, in fact, at
low temperatures the changing bond strength is the only temperature dependent para-
156
meter then a chain length distribution derived at about -25°C should exactly in-
clude the (narrower) distributions at still lower temperatures. Vice versa from one
measured step-strain free-radical curve all narrower ones must be predictable. John-
sen and Klinkenberg [11, 13] have confirmed that the chain length distributions
obtained at temperatures between -67 and -25°C are derived from one and the
same "master" distribution which is skewed to a different extent at the right hand
side. At higher temperatures (-4 and 17°C) discrepancies between the apparent dis-
tributions and the master distribution are observed. They are interpreted as indica-
tion of structural changes leading to a relaxation of some overstressed chains. These
chains may, however, break at a later stage of the loading process [11, 13]. The re-
laxation and slip effect is stronger at higher temperatures and at 50°C it clearlyout-
weighs the effect of diminished bond strength, i.e. the total number of broken bonds
at fracture decreases. Beyond 50°C most of the chain segments belonging to the
master distribution escape chain scission because the crystal blocks are no longer
able to exert and maintain large enough molecular stresses. At this point it may re-
main open which mechanisms are responsible for the stress relieving structural
changes.
In the above sections were discussed the scission of tie chains in highly oriented
fibers. Judging from the absence of any scale effect the molecular fracture events
are evenly distributed throughout the amorphous volume of the stressed sample.
Those "volume" concentrations of 1015 to 10 18 spins/cm 3 accounted for almost
the total radical intensity whereas the fmal fracture surfaces (with estimated radical
populations of 1012 to 10 13 spins/cm 2 ) did not noticeably contribute towards the
total number of radicals observed in tensile straining of the fiber samples. This is
different if the mechanical degradation is achieved through milling, grinding and/or
cutting. In all these techniques comminution is achieved through failure of virgin
material resulting from the application of multiaxial stress particularly involving
compressive components. In impact loading of platelike or granular samples com-
pressive waves may be generated. Upon reflection on boundaries or discontinuities,
compressive waves become dilational waves giving rise to tensile fracture. All forces
combine to overcome the cohesive forces and to create new surface area. Those chains
which break in this process are either intersecting weak cross-sections and/or are con-
tained in shear zones along a suitable direction. It must be assumed, therefore, that
chains are broken preferentially - but not exclusively - in the zones of new surface.
The surface concentration of free radicals in ground polymers has been deter-
mined by different investigators [20-23). As to be expected the mobility of the
chains within the degrading material strongly effects the obtainable radical concen-
trations. Backman and DeVries [21,22] sliced 66 PA, PP, and PE in a nitrogen at-
mosphere and transferred the slices within less than 0.5 s into liquid nitrogen and
determined the number of free radicals by ESR. The degradation temperature was
varied between -20°C and 90°C. The authors observed a rapid decrease of surface
concentrations at the respective glass transitions (Fig. 7.15). It should be noted that
the rate of decay of the radicals formed also increases beyond Tg' but that this loss
of radicals amounts to less than 30% [18]. From the level of the low temperature
radical concentrations it can be concluded that the average fracture plane breaks be-
tween 5 and 10% of all chains intersecting the plane. Pazonyi et al. [22] investigated
157
1.8
0
1.6 o Polyamide
~
';" • Polypropylene
..,
E
1.4 • Polyethylene
~
co
:- 1.2 ~
"
~\
.~ 1.0
'5
~ 0.8
~
"'"
"C
~ 0.6
~
•
\ '"'~
.;:: 0.4
0.2
a
0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
relative temperature TlTg
Fig. 7.15. Surface densities of spins observed on comminuted polymers as a function of the
degradation temperature T plotted as fraction of the corresponding glass transition temperature
[201.
polyvinylchloride with differing contents of plasticizer (30, 40, 50, and 60% respec-
tively of dioctylphthalate, DOP). They estimated that cutting at 30°C led to the
scission of (128, 105, 55 and 34% respectively of all) chains intersecting the fracture
plane. They also investigated low and high density polyethylene and found 1.39 and
1.17 . 10 15 spinS/cm 2 which means they had to cut 120 and 104% respectively of the
intersecting chains. Pazonyi et al. did their cutting in an ethanolic solution of diphenyl
picryl hydrazyl the color change of which was used as a measure for the number of
free radicals formed. The unusually large molecular cutting ratios given by Pazonyi
et al. and their discrepancy with the values cited earlier are noted and discussed by
Salloum and Eckert [23]. They point out that reactions of DPPH with ethanol and
with polymer surfaces not containing free radicals may take place thus accounting
for the excessively high radical numbers determined by Pazonyi. Salloum and Eckert
drilled PE, Nylon, and PP in a solution of DPPH in benzene. They obtained 0.1-2
(PE), 0.5-4 (6 PA) and 7-9 . 10 13 spins/cm 2 (PP) respectively. The first two values
are in agreement with the results of Backman and De Vries, the latter is not.
Both laboratories [20-22] conclude that free radicals are formed within a sur-
face layer of finite thickness. From a study of oxygen diffusion rates and of the con-
version of the original ESR signal from secondary 6 nylon radicals to that of peroxy
radicals Backman and DeVries determine that 50% of all "damage" occurred within
a layer of less than 0.6 JlIl1 from the surface. The remaining 50% of chain radicals
were found in a depth of up to 3 JlIl1 from the surface. Considering the morphology
of the degraded polymers, the mechanics of the comminution process, and the mo-
bility of the primary formed free radicals the spatial distribution of the secondary
radicals can be understood. In this case the pullout and breakage of individual PA
chains appears unlikely in view of the crystal strength. As we have discussed in Chap-
ter 5, a 6 PA chain embedded within a crystallite with more than 1.7 nm of its length
will rather break than be pulled away from the crystallite. The pull out of whole
158
microfibrils from a fracture surface very probably will occur and will be accompanied
by breakage of interfibrillar tie chains thus introducing damage within a surface layer
of up to 1 pm depth. Even more important will be the heavy plastic deformation of
the material ahead of a growing fracture surface. The propagation of free radicals
certainly contributes to deepen the layer of apparent damage. The layer depths ob-
served in these experiments do coincide, however, very well with the lower limits of
particle sizes obtainable in mechanical degradation [24] indicating, that the damage
may indeed be introduced mechanically up to the observed depths.
D. Effect of Sample Treatment
The effect of heat treatment on structure, fracture strength and chain scission has
been very extensively investigated by Statton, Park and DeVries [25-27] and by
Uoyd [5). With respect to our understanding of chain breakages in annealed fibers
the following morphological changes seem to be particularly noteworthy. Relaxed
heat treatments result in [25]:
1. Increases of the degree of crystallinity, perfection of crystallites, density, chain
folding, Young's modulus, and intensity of the fluid-like component of the NMR-
signal.
2. Decreases of sample length (shrinkage), breaking strength, sonic modulus, and Tg'
The changes are the more intense the higher the annealing temperature. Tension
during annealing tends to modify or reduce the above listed effects. Relaxed and
tension annealing do, however, increase the long period.
The morphological changes certainly have a bearing on the molecular stress dis-
tribution and chain breakage. In Figure 7.16 the histogram from an ESR step-strain
test is reproduced which was obtained from a tension-free annealed polyamide sample
at 199°C [27). The histogram is broader, indicating a less homogeneous distribution
of molecular strains, and it is shifted towards higher strains by about two to four
100
"'E
o
,;;-
'S? 50
III
.5
5}
<I
Fig. 7.16. Histogram of radical concentra-

tion from 6 polyamide fibers slack annealed
5 10 20 at 199°C and subsequently step-strained at
strain 0'0 room temperature [4. 51.
159
per cent of strain. This may be seen from Figure 7.17 where the chain length distri-
bution of the annealed sample is compared with that of an unannealed reference
sample. The used probability plot had been calculated under the assumption that
the histogram in Figure 7.16 represents a normal distribution of chain lengths skewed
at the right hand side.
99.99'r----------------,
%
98
90
50
10
Fig. 7.17. Probability plot of histogram of

0.01 .............1.....--1..........1_-'---'-----'-_.1...-...1.....--1..........1 radical concentration of 1. reference sample
89m n ~ D M ffi ffi % m
Strain and 2. sample slack annealed at 199 DC [4,51.
The observed conformational changes in slack annealing have been interpreted

[25-27] as an increase in relative length of (the formerly) extended tie chain seg-
men ts due to defect migration out of the crystal blocks. The number of regular
chain folds also increases. Shrinkage of the yarn appears to be governed by the num-
ber of folds introduced. The structural changes occurring during annealing are me-
chanically stable and not simply reverted under tensile stress. In Figure 7.18 a model
repres~ntation of the conformational changes in annealing are given [4, 5]. Because
of defect migration one obtains in tension annealing a relaxation of local molecular
strains and thus a larger number of extended chain segments. On the basis of this
experimental evidence Statton has constructed a string model of slack annealed,
tension annealed, and reference fibers which is shown in Figure 7.19 [28]. The effect
of the annealing process on the total number of free radicals formed in subsequent
straining at room temperature and on fracture strength is shown in Figures 7.20 and
7.21. The annealing at maximum tension (11.7% elongation) does lead to a smaller
number of spins at fracture and may be explained by the breakage of chains during
tension annealing. Intermediate positive annealing strains (not shown in Fig. 7.20)
do slightly increase the number of spins. The opposite effects of annealing tempera-
ture and of constraint of the sample during annealing on the homogeneity of the
chain length distribution are shown in Figure 7.22. These effects are discussed in
more detail in Chapter 8.
160
I
1 \~ l~\ ~ ,~ ~ ~ ~ n
a
chain end b sciss ion
c
Fig. 7.18. Model of tie chain conformation in (a) reference sample, (b) fiber slack annealed, and
(c) fiber annealed under tension [4,5 J.
SLACK ANN£AL HIGHEST TE NSION

CONTROL J\/'INEAL oll99'C
01199'C
Fig. 7.19. String-model by Statton [27,

28 J representing effect of annealing on
fiber morphology and chain conforma-
tion.
10
-CONTROL
0 ...,£)_0
-I:J----·::::tr:·~~ ,,::0 _.00'/. 0
o "'-------0--_
o 0 0
;;
'"c
SLAC~
~6 o 0
11 7 '/, <0
_ <0 -
__CD __ (] - -
Fig. 7.20. Concentration of free radicals at
120 160 180°C fracture (at room temperature) as function
QMeoilng temperature of annealing temperature [27 J.
161
1000
117~0~
~ ..---0
TI~
_ 0./"0"--- 0 0 0,
o9oo\@~25 CD 0 CD CD
~ ~-D_ CD --CD
'" CONTROL ~ SLACK
o 0
"b
~
800
Fig. 7.21 Effect of annealing temperature

120 140 160 180 and annealing condition on room temperature
annealing temperature fracture stress of 6 polyamide fibers [271.
0.9r-----------~----
117-
,-'
0
IT ~
Fig. 7.22. Effect of annealing temperature and con-

strain t during annealing on the strained segment
SLACK homogeneity 1T which is defined as the ratio of that
0.7 1'-,2""0--1~40'-----16~0----".->.J strain where the first radicals are 0 bserved to the strain
annealing temperature at break [271.
II. Free Radicals in Stressed Rubbers
A. Preorientation, Ductility, and Chain Scission
In Section I we have discussed the stress-induced chain scission of thermoplastic

polymers. As we have seen chain scission could occur whenever the intermolecular
forces acting on tightly embedded sections of extended (tie) molecules were strong
enough to resist segment slip during deformation in such a manner as to give rise to
axial forces equal to or larger than the mechanical strength of the molecular chain.
Exactly the same considerations apply to chain scission in elastomers which are in
a temperature region below Tg • Andrews, Reed, and Natarajan [29-31] have studied
162
under these aspects the deformation behavior of sulphur cured natural rubber (cis-
polyisoprene), dicumyl peroxide-cured natural rubber, and polychloroprene. They
observed that (see Fig. 7.23) in the temperature region I from glass transition (200K)
to 160 K the deformation behavior of natural rubber was very similar to that of a
thermoplastic polymer, with a clearly discernible yield point and a yield stress which
6
m. f'
1\ II.
\
\ I
I
I.
\ Z
Qz I
I \ \
I \ \ ..... 0 I
I \ \
\ iiiCi I
I \ zw
\ \ <~ I
I
e: I
,,
CRAZE I \ BRITTLE \ SHEAR
I
FORMATION ts \ FRACTURe YIELDING
I
II 51§
z \
,, , I
I
Zw I I
I I
I ~~ I I
I
..... I I
I I I I
1
100 120 140 160 180 200K
TEMPERATURE
Fig. 7.23. Deformation and fracture behavior of preoriented polyisoprenes and polychloroprene
(after Andrews and Reed, [31 J.
increased with decreasing temperature. Following a transition region there was a

narrow temperature band (II) between 150 and 130 K within which brittle fracture
occurred at strains of less than 5%. For still lower temperatures - and only for these
temperatures - the authors [29-31] made the very interesting observation that the
deformation behavior was determined by the state or orientation of the stressed net-
work. Unoriented rubber specimens behaved brittle just as they do within the tem-
perature band above. A second group of samples was prepared by first preorienting
them at room temperature by uniaxial straining to 100 to 200%, and then cooling
them down to 90 to 130 K (region III). These samples showed a ductile deforma-
tion behavior (Fig. 7.23). In this case the ductile low temperature deformation was
accompanied by microvoiding in narrow striations, the formation of detectable
amounts of free radicals, and by absorption of environmental gases within the micro-
voids. These gases expanded upon warming of the deformed specimens and gave rise
to a peculiar foaming effect. The observed free radical spectra of sulphur-cured poly-
isoprene have been assigned to allyl main chain radicals stemming from the scission
of the (weakest) bond located between the two a-methylene groups. These radicals
appeared together with a contribution of possibly RS radicals. The spectra of dicumyl
peroxide-cured polyisoprene and polychloroprene were assigned to secondary peroxy
radicals ROO. The concentration offree radicals (in polychloroprene) varied with
strain rate and yielded a maximum of 6 . 10 16 spins/g at a strain rate of 0.01 S-1
[31 ].
Taking all these observations together, the following model of the deformation
behavior of elastomers below their glass transition temperature is proposed: In region I
intermolecular attraction is sufficiently strong to permit energy elastic deformation
of chain segments. Gradually at first, and exclusively beyond the yield point, slip and
163
reorientation of chain segments occur. Chain scission is negligible because chains will
rather slip than break. In temperature region II where brittle fracture occurs irrespec-
tive of the state of preorientation the intermolecular attraction seems to be just large
enough to permit the axial loading of chain segments up to their breaking stress. In
the absence of local strain hardening the largest crack initiated within the specimen
during deformation to the 5% level will rapidly expand and prevent the growth of
any other crack nuclei. As discussed for thermoplastics, the formation of essentially
~ne plane of fracture is hardly sufficient to produce a detectable amount of free radicals.
Unoriented rubber specimens (samples frozen in the relaxed state) show the same brittle
behavior also in temperature region III (90 to 130 K). Specimens preoriented by ex-
tensions of 100 to 200%, however, obviously have gained the property oflocal strain
hardening. The nature of this effect can only be speculated on and probably is re-
lated to the build up of partially oriented microfibrils within the preoriented elasto-
mers. The result of local strain hardening is that the flaws and defects which are still
present develop into a system of interconnected microvoids and extended fibrils form-
ing the large and clearly visible striations. Environmental gases are adsorbed to the
newly formed surfaces. Since the micro fibrils are strong enough to carry the load
transferred upon them the strained sample will not break but more and more defects
will develop into crazes and striations. The largest number of striations appears at
medium extension rates (0.01 s-1 in polychloroprene). The large local deformation
(up to more than 100% strain) which is found throughout the total volume of the
sample together with the strong intermolecular attraction between different chain
segments leads to a large axial excitation - and breakage - of those chain segments
which play the role of tie molecules. As the discussed results show, the spin concen-
tration at break of elastomers in temperature region III is comparable to that ob-
tained from 6 polyamide. In case of a large preorientation of 300% (in polychloro-
prene) a ductile deformation behavior and the formation of free radicals are observed
[31]. The deformation is macroscopically homogeneous. No striations or crazes are
visible indicating that the strain hardening is more effective than at lesser preorien-
tations. The growth of microvoids obviously is stopped before they coalesce and
form crazes. The absence of large scale gas adsorption and of foaming upon subse-
quent heating together with differences in the observed ESR spectrum support this
view [31].
Brown, DeVries, and Williams duplicated these experiments and extended them
to Hycar 1043, an acrylonitrile-butadiene rubber [32] and to Silastic E RTV, a silicone
elastomer [33]. They confirmed for these polymers the above described effect of pre-
strain on the stress-strain behavior at low temperatures (118 to 193 K) the formation
of free radicals with increasing sample strain, and the effect of strain rate on radical
concentration with a maximum at a rate of 9 . 10- 4 S-1 . For these polymers they
did not observe, however, the three temperature zones mentioned above which are
characterized by distinctly different types of mechanical response. They also propose
that during preorientation numerous regions of higher strength and degree of orien-
tation are created which later serve to arrest developing microcracks before those
can attain a critical size. The fact that for the latter two polymers [32,33] only one
region of ductile and one region of brittle fracture were observed may be attributed
to different physical and chemical properties of those networks (configuration, cross-
164
link density, glass transition temperature). Some of these structural parameters will
be studied in the following section.
B. Cross-Link Density, Impurities, Fillers
We have seen above that in low temperature ductile deformation of rubbers a large
number of chain segments break which indicates that the local axial stresses exceed
the strengths of the chain segments. We have shown earlier that the stresses required
for bond breakage are more than two orders of magnitude larger than those applied
macroscopically to deform the rubber specimens. The appearance of free radicals,
therefore, indicates the highly inhomogeneous distribution of local stresses. The
radical intensity gives an account of the number N of segments which have been
stressed beyond their strength; N will depend, therefore, on all parameters which
raise the local stresses like cross-link density or content of reinforcing fillers, which
decrease the bond strength like presence of hetero-groups in the main chain, or which
affect the number of observed radicals through radical reactions.
Andrews and Reed [31] observed an increase in radical intensity with increasing
cross-link density of SUlphur cured natural rubber (Fig. 7.24) using the preorienta-
tion-technique described above. This observation is well in keeping with the fact that
12
dicumyl peroxide-X
non-purified
10
X- synthetic polyisoprene
dicumyl peroxide
~ 6 X-dlcumyl peroxide - purified

~
0::
>-
Z X-dlcumyl perox Ide - purified
~ 4 & deproteinized
z
o
u
z
f}j 2
Fig. 7.24. Effect of crosslink
density and crosslinking agent
OL-________________~----------~----~ on the concentration of free
a 2 3 5 6
radicals at fracture at 100 to
DEGREE OF CROSS LINKING 120 K (after Andrews and
('/. S for sulphur cure I Reed [311).
tensile stresses of identically strained rubbery specimens increase with increasing

cross-link density i.e. decreasing chain length between cross-links. The effect of im-
purities on the concentration of formed free radicals although clearly demonstrated
by the data of Figure 7.24 is not fully understood. In the absence of conspicuous
165
radical reactions or a direct effect on bond strengths it is believed that the impurities
constitute stress raising inhomogeneities [31].
It is not always necessary to prestrain rubbers to induce the development of a
large number of microcracks distributed throughout the bulk of the rubber sample.
Wilde et al. [34, 35] have shown that certain granular fIllers can promote bond rup-
ture during sample fracture. Again the studies were by necessity conducted at tem-
peratures below 210 K where the radicals were sufficiently stable. Fillers in elasto-
mers are commonly used for a variety of reasons. In any case the presence of such
fillers increases the complexity of the micromechanics of deformation and fracture.
The presence of the filler results in very non-homogeneous localized stress fields
ranging from levels of high strain concentration to complete separation between the
matrix and filler (dewetting). The filler particles may serve as crack arrestors and/or
initiators and thereby alter the fracture strength, resilience, and/or toughness. Wilde
studied four systems all filled to approximately 50% by weight. These were:
1. 29 J.1.ID glass beads in EPDM,
2. silane treated 15 fJ.m glass bead filled polyisoprene,
3. NaCI fIlled polyisoprene and
4. HiSil 233 (silica) filled polyisoprene.
The first system provided very weak interaction between filler and matrix, the
last very strong interaction, and the other two demonstrated intermediate filler-
matrix interaction. All systems were tested in temperature region I (I 70 to 200 K),
a region within which an unfilled rubber without pre orientation produces no free
radicals in tensile straining. Three of the four filled systems, however, behaved dif-
ferently. It was found that the untreated glass filled elastomer fractured with no
detectable production of free radicals, fracture in the rubbers filled with silane
treated glass and NaCI produced easily detectable concentrations of radicals
(3.21 . 10 14 spins/cm 3 ). The HiSil system produced at some temperatures and
strain rates, up to an order of magnitude more radicals than NaCI or treated glass
(7.86· 10 14 spins/cm 3 ). In his PhD thesis Wilde [35] details a rather comprehensive
comparison of scanning electron microscope photographs with the ESR results. The
photographs indicate that at room temperature dewetting started to occur in the
untreated glass system at strains of less than 10 to 20%, in the treated glass beads
and NaCI systems it occurred at strains of between 50 and 100%, and in the HiSil
system at strains greater than 200%. Scanning electron micrographs of the fracture
surfaces at all temperatures between 150 and 300 K revealed the same findings and
showed 1. that untreated glass beads were sitting essentially loose in the "smooth"
vacuoles or voids on the fracture surface, 2. that treated glass surfaces and NaCI
behaved similar to untreated glass except that here the vacuoles were not so smooth
and that there was material adhering to the filler particles, and 3. that the HiSil
particles were embedded in the matrix at the fracture surface with the rubber ad-
hering completely to the particle surfaces.
A slightly different two-phase system with strong bonds at phase boundaries
is that obtained from triblock copolymers such as styrene-butadiene. As discussed
in Chapter 2 such a copolymer molecule is composed of rigid (styrene) end blocks
joined by elastomeric (butadiene) center blocks. The styrene blocks aggregate to
form small domains which act as cross-links leading to a rubberlike elasticity of the
166
III. Mechanically Relevant Radical Reactions
block copolymer at ambient temperatures but to the possibility of plastic deforma-

tion at high temperatures. For an understanding of the fracture behavior of these
systems it would be useful to identify in which of the phases molecular rupture
most often occurs. The direct means of accomplishing this goal would be to frac-
ture the material and analyze the hyperfine structure of the resulting ESR spectra.
In the temperature range between liqUid nitrogen and room temperature, however,
tensile straining does not lead to sufficiently observable accumulations of free radicals.
As a consequence, De Vries, Roylance, and Williams [36] adapted the less conclusive
but more straight-forward method of comparing the spectrum of styrene-butadiene
block copolymers (SBS) with those of styrene and butadiene alone. These studies
were done on grindings to accumulate large amounts of fracture surface and at liquid
nitrogen temperatures. The low temperature made the radicals more stable and hope-
fully "froze" the primary radicals. Comparison of the spectra of the three materials
showed that the spectrum in the SBS contained all the lines of the butadiene radical
but not of the styrene radical. The radical of the SBS system was, therefore, assigned
to the butadiene phase. Unfortunately the study leaves open the questions as to
whether or not the radical formed by grinding at low temperatures is the same as
that presumably formed under normal conditions at room temperature and whether
the observed radical is the primary radical or a secondary one.
Taking together the experimentally observed effects of the five parameters im-
purity content, degree of cross-linking, ease of crystallite formation, strength of
filler-matrix adhesion, and presence of different phases on the nature and intensity of
free radicals formed we derive at the following conclusions. All five parameters tend
to increase the apparent cross-link density and decrease the extensibility of chain
segments between cross-links. They thus enhance the forces acting at a given strain
and also the ultimate forces that can be transmitted onto chain segments. Both con-
ditions are necessary before chain stresses can exceed chain strengths. Furthermore
all five parameters increase the number of actual or potential defect sites and cause
a better spatial distribution of chain scission events. If chains fail they will always
be located in the amorphous or more flexible regions of the (hetero-phase) elasto-
mers.
In Sections I and II of this Chapter primary and secondary free radicals have been
treated as microprobes which characterize occurrence and molecular environment
of chain breakages. As shown in Chapter 6 the primary mechano-radicals are always
chain end radicals which are mostly unstable. At a rate depending on temperature
these radicals will transfer the free electrons and thus "convert" to secondary ra-
dicals. This reaction and also subsequent conversion and decay reactions including
recombination are relevant with respect to an interpretation of the fracture process
in two ways. Firstly these reactions interfere with a determination of the concentra-
tion and molecular environment of the original chain scission points. Secondly they
change the physical properties of other network chains through the introduction of
167
unpaired electrons and the formation of crosslinks. For a discussion of spectro-

scopic details and of the stability and conformation of the free radicals the reader
is referred to the comprehensive monograph of R£nby and Rabek [37] and to the
review articles of Campbell [38] and of Sohma et al. [39]. In the following the me-
chanically relevant aspects of radical transfer and decay are reported.
A. Transfer Reactions
At this point the geometrical aspect of transfer reactions is to be investigated. Through

breakage of an elastically strained chain the primary radicals are formed at either side
of the breakage site. Because of the axial chain stresses the new chain end radicals
rapidly retract from each other. They are thus being prevented from recombination.
The primary radicals then convert to secondary ones mostly by transfer of a hydrogen
atom (see Table 6.1). If hydrogen transfer occurs along one and the same chain no
effect on the phYSical network properties is to be foreseen. If, however, a hydrogen
is abstracted from a neighboring chain the radical is transferred to a mid-chain posi-
tion at a hitherto uneffected chain segment. There is strong evidence for the occur-
rence of both processes, the radical migration along a chain [39] and the hydrogen
hopping accross chains [40]. The hydrogen hopping process was demonstrated by
Dole et al. [40] who had molecular deuterium gas, D2 , diffuse into irradiated poly-
ethylene; D2 was then converted to HD by the following reactions:
R + D2 --RID+D
b + R'H--R' +HD.
Although in this experiment a gaseous tracer had been used it is believed that
hydrogen abstraction from a hydrocarbon chain proceeds in the same manner. In
view of the mobility of the chain end radical the hydrogen abstraction seems to be
a logical vehicle for the transfer of a free electron to a central position within another
chain segment.
The radical migration along a chain was found by Sohma et al. [39] to give the
only plaUSible explanation for his observation that in PE and PMMA chain end radicals
convert to a near-end radical:
H H H H H
C - C' - - C - C C - H
H H H H
(77 K) (5 min. at 152 K)
CH 3 H CH3 H CH 3 H CH 3 H
C C C C' C C C CH.
R H R H R R H
(after short time (after 0.5 to 24 h.
milling at 77 K) milling at 77 K)
168
Sohma indicates that also in these polymers radical migration along the chain
and successive hydrogen abstraction are both important mechanisms of radical migra-
tion.
Transfer reactions are, of course, influenced by the presence of impurities or
additives which are active as radical acceptors or radical scavengers. Their mechanical
relevance lies in the fact that they reduce or prevent transfer of free radicals to cen-
tral chain positions, radical recombination, and/or cross-linking. The latter aspect is
discussed in Chapter 8. For a review of radical reactions with H2 , D2 , O2 , S02, NO
and N0 2 , NH 3 , H2 S, and various halogens the reader is referred to the works of
Campbell [38] and of RImby and Rabek [37], the latter being especially concerned
with the reactions of oxygen in its different forms (triplet and singlet molecular
oxygen, atomic oxygen).
B. Recombination and Decay
The normal decay of macroradicals by recombination in the solid phase is a second

order reaction [5,11,17-18,37-39,41-43]
[R(t)] = [Ro ]/(1 + k [Ro ]t). (7.10)
The decay rate term kRo depends - as expected - on temperature: for unloaded
fibers under He atmosphere values of 0.0014 S-1 (75°C) and 0.0110 S-1 (100 0c)
have been determined [18]. This value ties in with the 0.001 s-1 measured at 50
and 60°C under N2 [11,41]. It should be noted that the recombination reaction
is slowed down by the application of tensile forces as for instance during the pre-
fracture loading period: under near maximum tensile stress the decay rate terms
were only 0.00035 S-1 (75°C) and 0.00075 S-1 (100 0c) respectively [18]. The
kinetics of decay is also affected by conversion reactions (e.g. of alkyl into allyl
radicals) or catalytic effects [37].
The effect of the mobility of the radical site on the recombination rate has been
clearly shown by the pressure [44-46] and temperature dependence [41-43,47-49]
of radical decay in e.g. PE, PP, PA 6, 66, and 12, and PPO. The decay curves of free
radical populations in (irradiated) polymers as a function of increasing temperature
reveal a number of plateaus each corresponding to radicals trapped in a particular
morphological site. A transition from one plateau to the next indicates that one
species of the trapped free radicals has become sufficiently mobile for recombination
(chemically identical radicals in crystalline and amorphous regions may differ in steric
configuration which also effects the rate of recombination [37,42,47]). The transi-
tions are hence associated with the corresponding mechanical 'Y, (3, and a dispersions
[41-43).
As already indicated earlier apparent steps in those radical concentration curves
which are obtained by straining fibers by equal increments at different temperatures
[11,19], are not solely related to dispersion regions of the modulus but reflect de-
tails of the relative segment length distribution as well.
169
For chemically different radical species generally different decay constants with
different activation energies are observed. Tino et al. [42] determined for the central
polyamide radical-NH-CH-CH2- in the temperature region of 15 to 50 °c an
activation energy for decay of 10 kcal/mol, for the chain end radical-CONH-CH 2
a value of 5.2 kcal/mol.
Studies of the effect of hydrostatic pressure on free radical decay have the aim
of elucidating the mobility and the distance of migration. A series of experiments
has been carried out at the Polymer Institute of the Slovak Academy of Sciences in
Bratislava. The rate constants of free radical decay decrease exponentially with in-
crea~ing pressure [44-46]. At low temperatures the rate constant varies only slightly
with pressure. The retarding effect becomes stronger at higher temperatures; the
closer the decay temperature to the glass transition temperature Tg the more pro-
nounced the stabilizing effect of pressure. Of course the pressure effect shows a
saturation once the pressure is so large as to inhibit the molecular motion in question,
8 kb for the Q( relaxation in PE and PVAC at 80-110 0 C [44], 15 kb for the Q( relaxa-
tion in 6 PA [44J.
c. Anomalous Decay
Sohma et al. [39] investigated the thermal decay of mechano-radicals. They observed
for PE, PP, and PTFE an increase in the concentration of free radicals while increas-
ing the temperature from 77 to 170 K. They termed this behavior which is not found
for irradiation-produced radicals, anomalous. The anomalous increase is enhanced by
excess triboelectric charges of the sawed samples and by the presence of oxygen [39].
On the basis of their extensive investigations the authors advance a mechanism for
the creation of free radicals through heat treatment at the rather low temperatures
of 77 to 170 K. They postulate the formation of (spectroscopically neutral) carb-
anions during the degradation process at 77 K, here illustrated for PP:
-CH 2-(CH 3)CH + e- ~ -CH 2-(CH 3)CH-

t
(carbanions, no ESR spectrum)
+
-(CH3)CH-CH2 + e- ---. -(CH3)CH-CH2".
During the "heat" treatment (100-173 K) the carbanions react with O2 to form
spectroscopically active peroxy radicals. The released electrons leak to the ground:
-CH2-(CH 3)CH- + O2 + e-
~ -CH2 -(CH 3)HCOO
-(CH3)CH-CH2" + O2 ~ -(CH3)CH-H2COO + e-.
D. Radical Trapping Sites
Mechanically relevant information has been obtained from an analysis of radical

trapping sites. Thus it has been concluded in previous sections of this chapter that
170
it is the amorphous regions of a semicrystalline polymer where mechano-radicals

are formed. Apart from that changes in sample morphology during mechanical work-
ing have been investigated by ESR technique. Kusumoto, Takayanagi et al. [50-51]
studied PE and PP films by successively removing the amorphous material through
nitric acid etching. They then analyzed the ESR spectra obtained by 'Y irradiation
of the so treated films. Thus they were able to correlate the octet observed in PP with
radicals trapped at defect sites within the crystallites, the nonet with radicals in the
amorphous fold surfaces. The latter are especially effective radical trapping sites.
The authors analyzed the effect of quenching, annealing, and cold-drawing on the
mosaic-block structure of their fllms.
From a change in line shape parameter of the PE sextet spectrum and of the
temperature dependence of radical recombination rates the authors determined the
increase in unoriented material content during mechanical fatigue [51]. For the
study of the surfaces of as-grown and annealed PE single crystals Kusumoto et al.
[52] also employed the spin-probe method.
The ESR technique has permitted to identify initial rupture points in mechanical-
ly degraded copolymers. Lazar and SZQCS [53] investigated random, block, and graft
copolymers of methylmethacrylate (MMA) and styrene (S). They found in a degraded
mixture of equal parts MMA and S homopolymers equal concentrations of MMA
and S radicals. In a random copolymer (MMA : S = 1 : 1) the styrene radical prevailed
clearly. By comparing the reactions of mechano-radicals in ball-milled PP with those
of "'Y-radicals" Kurokawa et al. [54] concluded that the mechano-radicals were pro-
duced and trapped on the fresh surfaces created by the breaking-up of solid poly-
propylene.
1. S. E. Bresler, S. N. Zhurkov, E. N. Kazbekov, E. M. Saminskii, E. E. Tomashevskii: Zh.

tekhn. fiz. 29, 358 (1959); SOy. Phys.-Techn. Physics 4,321-325 (1959).
2. J. Becht, H. Fischer: Kolloid-Z. u. Z. Polymere 229,167 (1969).
3. G. S. P. Verma, A. Peterlin: Polymer Preprints 10,1051 (1969).
4. H. H. Kausch, K. L. De Vries: Int. Fracture 11,727-759 (1975).
5. B. A. Lloyd: Ph. D. Thesis, Department Mechanical Engng. Univers. of Utah (June 1972);
B. A. Lloyd, K. L. DeVries, M. L. Williams: 1. Polymer Sci. A-2, 10,1415-1445 (1972).
6. J. Becht: Dissertation, Technische Hochschule, Darmstadt (1971).
7. H. H. Kausch, J. Becht: Rheologica Acta 9, 137 (1970).
8. M. A. Gezalov, V. S. Kuksenko, A. I. Slutsker: Mechanica Polymerov 8,51 (1972);
Polymer Mechanics USSR 8,41 (1972).
9. B. Crist, A. Peterlin: Makromol. Chern. 171, 211-227 (1973).
10. W. H. Hassell: M. S. Thesis, Department Mechanical Engng. Univers. of Utah (June 1973).
11. U. Johnsen, D. Klinkenberg: Kolloid-Z. u. Z. Polymere 251, 843 (1973).
12. H. H. Kausch: J. Polymer Sci. C 32, 1-44 (1971).
13. D. Klinkenberg: 38. Dtsch. Physikertagung, Niirnberg, 23.-27. Sept. 1974. D. Klinkenberg,
Dissertation, Technische Hochschule, Darmstadt 1978.
14. M. Williams, K. L. De Vries: Proc. Fifth Internat. Congr. Rheology, S. Onogy, Ed., Vol. 3,
University of Tokyo Press, Tokyo 1970, 139.
15. B. Yu. Zaks, M. L. Lebedinskaya, V. N. Chalide: Vysokomol. soyedA 12,2669 (1970);
Polymer Sci. USSR 12,3025 (1971).
171
16. H. H. Kausch, J. Becht: Kolloid-Z. u. Z. Polymere 250, 1048 (1972).

17. J. Becht, H. Fischer: Kolloid-Z. u. Z. Polymere 240, 766 (1970).
18. K. L. DeVries, B. A. Lloyd: Univers. of Utah, personal communication.
19. L. A. Davis, C. A. Pampillo: 1. Polymer Sci., Polymer Phys. Ed. 11 ,841-854 (1973).
20. D. K. Beckman, K. L. DeVries: Department Mechanical Engng. Univers. of Utah,
June (1969).
21. D. K. Backman, K. L. De Vries: J. Polymer Sci. A-I, 7,2125 (1969).
22. T. Pazonyi, F. Tudos, M. Dimitrov: Angew. Makrom. Chern. 10, 75-82 (1970).
23. R. J. Salloum, R. E. Eckert: J. Appl. Polymer Sci. 17, 509-526 (1973).
24. T. Pazonyi, F. Tudos, M. Dimitrov: Plaste u. Kautschuk 16, 577 -581 (1969).
25. W. O. Statton: J. Polymer Sci., (Polymer Symposia) 32, 219 (1971).
26. J. B. Park: Ph. D. Thesis, Department Mechanical Engng. Univers. of Utah, (June 1972).
27. W. O. Statton, 1. B. Park: to be published J. Polymer Sci., Polymer Phys. Ed.
28. W. O. Statton: Univers. of Utah, personal communication.
29. E. H. Andrews, P. E. Reed: 1. Polymer Sci. B-5, 317 (1967).
30. R. Natarajan, P. E. Reed: 1. Polymer Sci. A-2, 10, 585 (1972).
31. E. H. Andrews, P. E. Reed in: Deformation and Fracture of High Polymers, H. H. Kausch,
J. A. Hassell, and R. l. laffee, Eds. New York: Plenum Press 1973, 259.
32. R. Brown, K. L. De Vries, M. L. Williams: Polymer Preprints 11/2,428 (1970).
33. R. T. Brown, K. L. DeVries, M. L. Williams: Polymer Letters 10,327 (1972).
34. K. L. DeVries, T. B. Wilde, M. L. Williams: J. Macromol. Sci-Phys. B7 (4), 633 (1973).
35. T. B. Wilde: Ph. D. Dissertation, Department of Mechanical Engng. Univers. of Utah (1970).
36. K. L. DeVries, D. K. Roylance, M. L. Williams: 1. Polymer Sci. A-2, 10, 599 (1972).
37. B. Rlinby, J. F. Rabek: ESR-Spectroscopy in Polymer Research, Heidelberg: Springer-
Verlag 1977.
38. D. Campbell: J. Polymer Sci., Part D (Macromol. Rev., Vol. 4) 91-181 (1970).
39. J. Sohma, M. Sakaguchi: Advances in Polymer Sci. 20, 109-158 (1976).
40. M. Dole, D. R. Johnson, S. Kubo, W. Y. Wen: Department of Chemistry, Baylor Uni-
versity, Waco, Texas.
41. F. Szocs, 1. Becht, H. Fischer: Europ. Polymer 1. 7, 173-179 (1971).
42. J. Tino, J. Placek, F. Szocs: Europ. Polymer 1. 11, 609-611 (1975).
43. T. Nagamura, N. Kusumoto, M. Takayanagi: J. Polymer Sci. (Polymer Phys. Ed.) 11,
2357 -2369 (1973).
44. F. Szocs, 1. Tino, J. Placek: Europ. Polymer 1. 9,251-255 (1973).
45. F. Szocs, O. Rostasova, 1. Tino, J. Placek: Europ. Polymer 1.10, 725-727 (1974).
46. F. Szocs, O. Rostasova: 1. Appl. Polymer Sci. 18, 2529-2532 (1974).
47. H. Kashiwabara, S. Shimada, J. Sohma: ESR Applications to Polymer Research
(Nobel Symposium 22), Stockholm: Almqvist & Wikselll973.
48. T. Nagamura, M. Takayanagi: 1. Polymer Sci. (Polymer Phys. Ed.) 13, 567-578 (1975).
49. J. Tino, J. Placek, P. Slama, E. Borsig: Polymer 19/2,212--214 (1977).
50. N. Kusumoto, K. Matsumoto, M. Takayanagi: J. of Appl. Polymer Sci. A-I, 7, 1773-1787
(1969).
51. T. Nagamura, N. Kusumoto, M. Takayanagi: Intern. Conf. on Mech. Behavior of Materials,
Kyoto, Japan, Aug. 1971.
52. N. Kusumoto, M. Yonezawa, Y. M6tozato: Polymer 15/12,793-798 (1974).
53. M. Lazar, F. Szocs: J. of Polymer Sci. Part C 16, 461-468 (1967).
54. N. Kurokawa, M. Sakaguchi, J. Sohma: Polymer 1.10/1,93-98 (1978).
172
Chapter 8
The Role of Chain Scission in Homogeneous

Deformation and Fracture
I. Small-Strain Deformation and Fracture of Highly Oriented Polymers 174

A. Underlying Problems . 174
B. Loading of Chains before Scission . 175
C. Spatially Homogeneously Distributed Chain Scissions 183
D. Formation of Microcracks 193
E. Energy Release in Chain Scission 196
F. Fatigue Fracture of Fibers 198
G. Fractography . 200
II. Deformation, Creep, and Fatigue of Unoriented Polymers 204
A. Impact Loading 204
B. Failure under Constant Load. 211
C. Homogeneous Fatigue 220
1. Phenomenology and Experimental Parameters 220
2. Thermal Fatigue Failure . 222
3. Wohler Curves . 223
4. Molecular Interpretations of Polymer Fatigue. 224
D. Yielding, Necking, Drawing 230
E. Elastomers . 235
III. Environmental Degradation . 237
Section II B of Chapter 2 gave a description of the uniaxial deformation behavior

of an unoriented thermoplastic polymer. It was indicated that - depending on
experimental and material parameters - failure could occur at any of the different
stages of a tensile loading process:
as brittle fracture due to crack initiation and propagation during anelastic de-
formation
as a consequence of plastic flow after necking or homogeneous yielding
as ultimate failure following plastic deformation with strain hardening.
It is useful, therefore, to distinguish between brittle strength, yield strength,
and ultimate strength of a polymer.
Quite obviously the role of chain loading and scission within the three strength
determining mechanisms will be quite different. It has been stated repeatedly
throughout this book that the load carrying capabilities of chain molecules are uti-
lized most effectively if chain orientation and intermolecular attraction permit
the gradual accumulation of large axial chain stresses and oppose slip and
173
8. The Role of Chain Scission in Homogeneous Deformation and Fracture
void formation. It is for this reason that highly oriented polymer fibers are the most
suitable objects to study chain loading and scission. In Chapter 7 the experimental
observations concerning the formation of mechano-radicals and of their conversion
have been reported. With respect to the phenomenology there is little disagree-
ment in the literature. In the first section of this chapter the possible effects of
chain scission and of radical reactions on ultimate strength will be discussed, which
is a more controversical subject.
The second section concerns unoriented polymers and failure mechanisms, which
are homogeneous on a macroscopic scale. Thus the conditions of impact loading
(large loading rates), creep (constant load), yielding, necking, and drawing (constant
strain rate), and fatigue (intermittent or cyclic loading) will be investigated. Elasto-
mers also belong to the group of unoriented polymers although their failure may
occur after considerable elongation and orientation.
The chapter closes with a discussion of chain scission caused or assisted by the
simultaneous action of stress and aggressive environment.
I. Small-Strain Deformation and Fracture of Highly Oriented Polymers
A. Underlying Problems
The undisputed fact of the rupture 0/ chains in large numbers during mechanical
action [1] is, in itself; neither proof nor even indication that macroscopic stress relaxa-
tion, deformation and fracture are a consequence of the breakage of chains. As was
pointed out by Kausch and Becht [2] the observed total number of broken chains
is much too small to account (by virtue of their load carrying capability) for the
measured reductions in macroscopic stress. As shown in Figure 7.4 the stress relaxa-
tion within a strain step (of 0.65%) amounts to 60 to 100 MN/m 2 • The load having
been carried by the 0.7 . 10 17 chain segments broken during such strain step, how-
ever, is calculated to be 2.4 MN/m 2 if the tie-segments had traversed just one amor-
phous region; it is n . 2.4 MN/m 2 if they had spanned n such regions. For these and
most of the following calculations a sandwich-type model of the fiber structure will
be used such as represented in Figure 7.5 I. In the case of n = 1 the macroscopic stress
relaxation obviou~ly has an intensity 25 to 40 times larger than the decrease of the
accumulated molecular stresses derived from the number of the observed chain
scission events. However, here also one must say that this discrepancy in itself is
neither proo/nor even indication that macroscopic stress relaxation, deformation
and fracture are not a consequence of the breakage of chains.
A basic problem of microstructural fracture theories involving chain scission
is clearly an elucidation of this descrepancy. Such a clarification can only be at-
tempted by widening the scope and by including other experimental observations.
Three principal questions must be raised before supporting or discarding a chain
scission fracture hypothesis:
- Does the number of chain scission events observed by anyone method corre-
spond to the number of chains actually breaking?
174
1. Small-Strain Deformation and Fracture of Highly Oriented Polymers
Does chain scission effect the mechanical properties of a polymer network in an

indirect way, e.g. through the release of stored elastic energy or ensuing radical
reactions?
- At what point and why does the phase of spatially homogeneous defect devel-
opment turn into local crack growth? Is chain scission a necessary or all acci-
dental precursor to mechanical instability?
The discussion of the underlying problem of the nature of the primary molec-
ular processes of damage, their interaction and possible propagation has been initiated
and constantly stimulated through the various detailed studies carried out in the
A. F. Ioffe Physico technical Institute in Leningrad by S. N. Zhurkov and his col-
leagues [3-33]. The individual problems attacked concern
investigation of the loading of chain segments before chain scission by IR tech-
nique [4-16]
determination of the absolute numbers [16-18] of chain scission events using
ESR [19-21] and IR [22-26] spectroscopy
formation of microcracks studied by X-ray diffraction [27-29]
- kinetics of thermal, mechanical and molecular degradation [30-33]
analysis of the structure, mechanical behavior and fatigue under various environ-
mental conditions in view of the foregOing considerations. These are reviewed in
[3] and will be discussed later in this Section.
A general remark must be made concerning the comparison and transfer of
results obtained from different specimens (e.g. films or fibers), methods (e.g. IR or
ESR), or laboratories. Whereas relative data such as the effect of time, temperature
or stress on the behavior of one specimen (in one spectrometer at one laboratory)
are generally in good agreement inconsistencies still persist with respect to absolute
figures and to the possible structural and/or chemical identity of different specimens.
The following discussion of the fracture work and of fracture theories involving
chain scission will be centered around the three principal questions raised earlier
even though they will not always be answered.
B. Loading of Chains before Scission
To date two methods have been employed for the determination of the distribution
N (t/I) of (large) axial chain stresses:
the indirect method of observing the formation of free radicals after a lowering
of chain strengths through increasing temperatures
the direct method of evaluating the stress-induced shift of infrared (IR) adsorp-
tion bands.
The indirect method of obtaining stress-distributions is based on the fact that
the strength of a chain segment decreases with increasing temperature. During an
upward temperature scan, therefore, those chain segments will be broken successive-
ly whose strength t/lb(T) becomes smaller than their axial load. The method has
been discussed in Section ICI of Chapter 7 for groups of chain segments with dif-
ferent relative lengths L/Lo . The ensuing relative length distribution N (L/Lo) can
175
-..J
?O
0'\ '""'l
go
S'
o
'...,"
(')
5000 4000 3000 25 00 2000 1800 1600 1400 1200 1000 950 900 850 800 750 Sf
5'
! CI)
II @.
I I .1 to. I~ J l\ I ~ J 100.. ... -""" 1M . _.. '"o·
I .. rio iIIl ~ ::l
~ J~ I P"'"
, [\, A
r ~ ,,,.... }tI ,\,,1 5'
I _~; ~\J 'ilJ 1\l ~ ~~ f'; I~ ~ IL. .... ~ ~
~ i ...
, r' ! I
I" I
I
I
i IV
,
'V """.. ::r::
o
:3
I I 1 ~
(1)
::l
,I (1)
, "- ' t ... ..... _.- ! I ...+- ... o

! ! . I ~
! 1 o
- -I ; 1 "T' 8>
. .
_- - --1_.. .,3
g.
1 I I ::l
, I .,
5-
.,~
! ~
IDiNSI !:i
(1)
2 3 4 5 6 7 8 9 10 11 12 13
Fig. 8.1. Dichroism of polypropylene;

parallel dichroic vibrations at 842,975,998,1045,1168,1254,1304 cm- I
perpendicular dichroic vibrations at 809, 899, 941,1330 em-I.
easily be transformed into a stress distribution N [l/l (L/Lo)] with the aid of
Eq. (7.1).
This indirect method has a rather restricted applicability. It is limited to a tem-
perature region T < T 1 where interfibrillar slip is still negligible and to molecular
stresses l/l > l/lb(T 1)' Furthermore the method does not permit the solution of a
problem inherent to all ESR measurements; namely to determine whether the abso-
lute concentration of broken bonds is equal to twice the observed concentration of
free radicals or whether it is (much) larger.
To overcome this difficulty Zhurkov and his collaborators began in 1965 to
study by an independant method, the IR technique, the effect of axial stress on
molecular chains [4-16]. Subsequently similar [35, 37-40] infra-red studies were
carried out in the USA. All these investigations are based on the fact that frequency
and intensity of skeletal or backbone vibrations of chain segments respond to a
superimposed deformation of the vibrating skeleton. In polypropylene for example
C-C stretch vibration modes have been assigned [16, 36,41,42] to the parallel
dichroic bands (Fig. 8.1) at 1326, 1168, 1045,975,842,456, and 398 cm- 1 . All
of these bands have been found to be stress sensitive [36]. The stress effect manifests
itself as a shift of the frequency of maximum absorption and in the occurrence of a
"tail" at longer wavelengths (Fig. 8.2).
Frequency
E
'"::
.="'
Fig. 8.2. Schematic effect of stress on

a skeletal vibration of a semicrystalline,
highly oriented polymer after ([5, 361).
The mechanisms of the distortion of IR absorption bands of stressed polymers

have been discussed especially by Gubanov [7-9], Kosobukin [13], Vettegren and
Novak [15] and Wool [36]. There is a general agreement that the distorted IR absorp-
tion band profile D(v) can be related to a large number of separate oscillators with
strongly overlapping absorption bands whose maxima are shifted by different amounts.
The possible causes of the frequency shift of an individual stressed oscillator have
been considered to be the quasielastic deformation of the harmonic oscillator (re-
177
duction of force constant under stress), the elastic increase of bond angles, changes
of segments conformation, and the creation of defects. For small deformations one
predicts [4-16, 36], that for the first three mechanisms there is a quite linear change
of frequency with molecu lar stress l/I:
!::.v' = vI/! - Vo = ex' l/I. (8.1)
As discussed previously molecular stresses are a function of molecular strains

which depend on sample morphology and chain orientation. Any real sample, there-
fore, contains a variety of differently stressed oscillators. The profile of the deformed
band, D(v), representing such a system of oscillators can be expressed by a convolu-
tion integral:
00
D(v)=J F(n)U(v-n)dn. (8.2)
Here F (n) is the distribution of stressed oscillators, U (v) the shape of the nor-
malized undeformed band, and n an integration variable. The distribution F (v) can
be obtained by deconvolution and transformed into a molecular stress distribution
by Eq. (8.1).
As already indicated and shown in Figure 8.2 the distribution of molecular
stresses in semi crystalline polymers contains a peak of apparently homogeneously
stressed chain segments - related to the crystalline regions - and a tail of very dif
ferently stressed chains - related to highly stressed tie segments. For a regular sand-
wich structure of crystalline regions having a modulus Ee and a chain axis orientation
described by cos () one observes a frequency shift !::.v of the crystalline peak:
!::.v= ex' Ek Gcos 2 () = exG. (8.3)

Ee
Thus the macroscopic stress sensitivity factor ex depends on sample-morphology
and orientation. As shown in Figure 8.3 quite different stress sensitivities have been
observed for identical bands in different samples [36]. In any real sample the mac-
roscopic stress sensitivity factor ex is an average value. For the 975 cm- 1 band of
polypropylene it ranges from 2 to 6 cm- 1 per GN/m 2 (Fig. 8.3). The 1168 cm- 1
em
-,
ROYLANCE
4
+'
.J:.
"' ZHURKOV
2 VETTEGREN
WOOL
Fig. 8.3. Stress-induced shift of the peak of the

400 600 800 MNj m' 975 cm -1 absorption band of polypropylene as
stress found by different investigators 16, 15,34,361.
178
band of the same polymer has a larger stress sensitivity of 10 cm- 1 per GN/m 2
[15]. Vettegren and Novak [15,16] report an a of 2.2 cm- 1 per GN/m 2 for the
930 cm- 1 band of PA 6 and of 3.6-6 cm- 1 per GN/m 2 for the 974 cm- 1 band
of PETP. For the latter band depending on sample treatment Mocherla and Statton
fmd stress sensitivity values of 12 to 20 cm- 1 per GN/m 2 [40]. More recently Vet-
tegren et al. [16] have obtained a stress sensitivity value of 8 cm- 1 per GN/m 2 at
1350 cm- 1 for polyacrylonitrile.
In the tail region of the "stressed 975 cm- 1 band" of PP frequency shifts of
up to 35 cm- 1 are discernible which correspond to molecular stresses up to 12 GN/m2 .
Although the significance and dynamics of "stressed-IR" investigations are not yet
understood in detail a number of interesting observations have been made:
- the integrated area of a deformed band (975 cm- 1 PP) remained constant with
a increasing from 0 to 700 MN/m 2 [5] indicating that no new oscillators had
been created under stress;
- the area fraction in that band related to overstressed segments increased with
stress to 15% at a = 400 MN/m 2 and 18% at 490 MN/m 2 [12]; for the 1170 cm- 1
band an area fraction of 6% is reported [16] together with 15% for PETP
(976 em-I), and 7% for PA 6 (930 em-I), all at a stress of 500 MN/m 2 ;
- it is interesting to note that the segment density l/q L for PP segments of 5 nm
length amounts to 56· 10 19 segments per cm 3 of which 15% or 8.5' 10 19
segments/cm 3 would be overstressed at a = 400 MN/m 2 ;
at a constant stress level this tail area is directly proportional to the content of
non-crystalline material, i.e. the overloaded segments seem to be contained in
the amorphous regions [5];
the tail regions of the deformed bands (930 cm- 1 PA 6; 970 cm- 1 PETP;
975 cm- 1 PP) show fairly well defined limits at the lower frequency end corre-
sponding to fairly well defined limits of sustained molecular stress"'; these limits
are determined [6, 15, 16] to be 11-13 GN/m 2 in PP, 15-20 GN/m 2 in PETP
(lavsan),8 GN/m 2 in PAN and 16 to 20 GN/m 2 in PA 6;
with some samples [12, 24, 25] these limits are already reached at moderate
macroscopic stresses smaller than 50% of sample strength; in other samples the
limits are only attained at a stress level close to the breaking stresses [16];
1/Im
N 10
~
Z
~
a b
o~----~----~----~~ O~----~----~----~~
o 300 600 900 G' o 300 600 900 d
[MN/tn 2 ] [MN/m 2 ]
Fig. 8.4. Maximum stress on molecular segment, I/J tn' as a function of applied uniaxial stress u for
PP (a) and PETP (b) for samples of different strengths ub; Ub indicated by ~ (after 116)). T = 26°C.
179
Figure 8.4 reproduces the results ofVettegren et al. [16] on the maximum stresses
on molecular segments for three morphologically different samples each ofPP
and PETP; the - non-specified - morphological differences do not seem to effect
the limiting stress value; in samples a) the largest molecular stress concentration
(l/Imla) is larger than in b) or c) but in all three samples it is apparently extremely
homogeneous as indicated by the fact that the maximum value of l/I m is only
reached at the maximum of a; this behavior ties in with the observation [5] that
the deformation of a band (975 cm- 1 PP) of a highly drawn sample (A. = 10) is
much less than that of a sample drawn to A = 5;
the observed molecular stress limit is also a function of temperature; in PETP
Zhurkov et al. [6] determined a decrease from 22 GN/m2 at 100 K to 15 GN/m2
at 400 K (on the basis that the observed molecular stress limits correspond to the
chain strengths l/Ib and using wo/kb = 48· 10 12 S-l as for PA 6 segments one
derives an activation volume of 11.7· 10- 6 m 3 /mol and an activation energy
Do = 316 kllmol for the PETP segments); from later experiments Vettegren
et al. [16] obtained activation volums of 13.2 . 10-6 m 3 /mol for PETP,
15.0· 10- 6 m 3 /mol for PA 6 and 15.6 .10- 6 m 3 /mol for PP;
the dichroism of a band permits the determination of the average degree of orien-
tation (cos 2 0) ofloaded chain segments; Zhurkov et al. [5] found for a PETP
with cos 2 0 = 0.75 that the most highly stressed segments approached cos 2 0 =
= 1, i.e. complete uniaxial orientation (Fig. 8.5);
the distribution of molecular stresses in PP varies during deformation (creep)
and stress relaxation as indicated by the different time dependent intensity
changes of fixed frequencies in the deformed part of the spectrum (dynamic
polarized IR studies [6, 35])
1.0
0.8
(!)
N
0.6
UI
...0
0.4
0.2
0
900 950 cni11000~
Fig. 8.5. Average orientation (cos2 8) of highly stressed PETP segments as a function of stress;
as abscissa the frequency shifts of the 974 cm- 1 band and the corresponding axial chain stresses
(using a stress sensitivity factor Cl< = 3.6 cm- 1 per GN/m 2) have been employed (after [51).
180
creep deformation tends to increase the highly loaded chain area of the 975 cm- 1
band in PP at the expense of intermediately stressed segments [6, 35] further
supporting the hypothesis that the band deformation is indeed due to a shift of
frequencies and not to a creation of new oscillators .
stress relaxation after shock loading of highly oriented PP showed an initial 10
to 70 s period of fast decay at practically constant degree of crystal orientation
[35] as evidenced by the absorption of the 899 cm -1 band which is a highly
dichroic, crystalline or "helix" band [43]; at later times a "steady state" relaxa-
tion takes place which is accompanied by a decrease of absorption of the
899 cm-1 band indicating a decrease of crystallinity or of helical regularity
[35];
in shock loading PP (e.g. up to a strain of 10.5% in less than 0.1 s) the largest
absorption at 955 cm -1 is observed after t = 69 s when considerable stress re-
laxation has occurred whereas in ramp loading at a rate of lO%/min the largest
absorption is attained at the maximum of stress at 10.5% strain; the largest
increase in intensity at 955 cm- 1 is (3.2 times) larger in shock loading than in
ramp loading [38]; the molecular stress transfer in this highly oriented PP, there-
fore, is a viscoelastic process involving the deformation of the amorphous regions
and the resistance to uncoiling of helices; Wool [39] has carried out a detailed
experimental and computational analysis of stress relaxation, dynamic IR be-
havior and bond rupture; he concluded that different stress sensitivity values
for crystalline regions (2.1 cm- 1 per GN/m 2 ) and individual chains (8 cm- 1
per GN/cm 2 ) should be employed, that the most highly stressed chains
(952 cm- 1 ) relax first thus contributing to the increase in intensity at higher
frequencies (e.g. 955 and 960 cm- 1 ), and that bond scission will be negligible
if its activation energy Uo is equal to or larger than 121 kllmol (29 kcal/mol);
with Uo = 105 kllmol very little chain scission occurs (affecting IR tail inten-·
sities by less than 0.3%), with 84 kllmol the majority and with Uo = 63 kllmol
(15 kcal/mol) all of the overloaded chains are expected to break; in his PP
material chain scission did not seem to contribute measurably to the relaxation
phenomena, this argument was backed by the observation that the absorption
bands characteristic for chain-end groups remained constant within 0.5% [39];
on the other hand in the PP material of Vettegren et al. [16] stress relaxation
at 100°C was accompanied by the measurable appearance of chain-end groups.
The dynamic IR behavior of PETP has recently been investigated by Mocherla
et al. [40]. Apart from the already discussed stress effects the authors observed the
existence of an initial region of small stresses (up to 70 MN/m2) within which stress
effects (on the 973 cm- 1 band of PETP) were absent. They concluded from ex-
tended studies of differently heat-treated films that this initial region coincides
with the range of elastic response of the sample; obviously small elastic stresses are
predominantly transmitted by secondary forces not giving rise to axial distortion
of the chains.
Summarizing the above results from this important new method one may say
that amorphous chain segments exist in appreciable numbers which carry up to about
twenty times the average stress. There are upper limits of molecular stress which in
the case of polyamide 6 seem to be given by the chain strength (21 GN/m 2 ). With
181
polypropylene the upper limit of molecular stresses in the highly, uniaxially oriented
material of Wool is due to the inception of a distortion of helical chain conforma-
tions and of crystal lamellae. No bond breakages could be traced.
In loading their oriented PP films Zhurkov, Vettegren et al. [6-16] observed,
however, that an initially present concentration of 1-10 . 10 18 carbonyl groups
per cm 3 increased [24]. They found this increase and the formation of other end
groups (see below) related to the decrease of the number of overloaded chains [16].
The Russian authors concluded from the equality of the activation energies of ther-
mal and mechanical destruction of their PP films (29 kcal/mol) with that of end-
group accumulation (30 kcal/mol) that the kinetics of all three processes is con-
trolled by the same molecular process, the thermomechanical scission of chain seg-
ments. According to the calculations of Wool, chain scission should practically never
occur in PP if Vo is equal to 29 kcal/mol. Those calculations, however, most prob-
ably refer to a constant temperature in the network of loaded and overloaded
chains. Consideration of the large strain energy release at chain breakage would lead
to the adjustment that local temperatures T are higher than the ambient temperature
To and that the effective chain strengths, lh(T), are smaller than lJib(To). It seems
likely, therefore, that the phenomenological observations do not contradict each
other especially taking into account possible structural differences of the PP samples
of the laboratories.
The upper limits of molecular stresses deduced for PETP(15-20 GN/m 2) and
PAN (8 GN/m 2 ) were also related to chain breakage by the Russian authors [16].
With these materials, however, no comparable results on scission products (end groups,
free radicals) have been reported. In these cases as with the PP material of Wool the
upper stress limit seems to derive from the limited resistance of the crystal lamellae
against distortion and break-up.
The latter conclusion is supported by the investigations of Becht and Kausch
[44-48] concerning the deformation of highly oriented semicrystalline fibers. In a
regular sandwich structure critical axial forces can be exerted onto tie segments if,
and only if, the crystal lamellae can withstand stresses up to the chain strength. Other-
wise crystal disintegration would precede chain scission. Using the spin probe tech-
nique, calorimetry and molecular weight measurements Becht [44-47] demon-
strated the strain effect on crystal integrity. He irradiated samples of highly oriented
6 PA, 12 PA, PP, PETP, and PE by I-Mev electrons at liquid nitrogen temperature.
Subsequently, all samples were heated to their glass-transition-temperature (or
above) for at least 5 minutes; thus, all radicals in the amorphous phase disap-
peared, and radicals remained within the crystallites only. These samples were then
strained within the cavity of an ESR spectrometer at room temperature.
The rate-determining steps of the radical-decay reactions in these cases are trans-
fer reactions and oxygen diffusion. They are both very strongly correlated with the
mobility of the radical-carrying molecular segments with respect to their surroundings,
Le., the crystal lattice. An increase in these rate constants, therefore, is a measure of
an increase in mobility, which may be due to a decrease in intermolecular attraction
and/or to the introduction of new crystal defects.
In Figure 8.6, the concentration of free radicals is plotted versus a reduced time
KoAt, where Ko is the rate of radical decay a t zero strain and At is the time elapsed
182
,.....
[R1
+_
.0 £=0°'°
Iii 50
----+
---: _12.5°'0
""
c '"3'
--;- S-PA
" ••• --~ --"-
15 %
1--0-
.....
.~ S·'o -!.!So,•
+-- --
:g -.,
! 25 r - 1--
17·'0··· ..
-+
c I
o
1---
I ••
......
11°'.
..
I
I
+-----.ISO,.. . .
~
is 22.5°'0 ... I
u 10 ~
Cu
• PP
~ •••
28°'° ••••• PE
5
•
-100 0 100 200 300 400 10 3 Ko At
Fig. 8.6. Effect of sample strain e on crystal distortion as evidenced by the rate of radical decay
in irradiated and annealed highly oriented fibers (after (481).
from the inception of sample straining. It should be noted that in PE, PP, PETP, and
12 PA, the application of sample strain has a twofold result: an immediate and ir-
reversible decrease of the (crystalline) radical population and a stress-dependent re-
versible increase in the rate of decay. In 6 PA neither of these two effects can be
observed. It was concluded from this experiment that in strained 6 PA, the tie mole-
cules fail before axial chain stresses are reached which are sufficient to distort the
crystal lamellae. In PE, PP, PETP, and 12 PA, however, forces can be transmitted
which clearly affect the crystal lamellae. It is logical to assume that these forces are
primarily transmitted by the tie segments. One would have to conclude that the
axial load bearing capability of microfibrils of PE, PP, PETP and 12 PAis limited
through the beginning crystal distortion. Thus chain strengths are not fully utilized
in these polymers.
Still another combination of chain and crystal properties seems to exist in the
PE films used by Vettegren et al. [15-18] for their end-group studies. Under stress
vinyl, methyl, and oxygen-containing end groups formed readily and in large con-
centrations while no stress effects on IR bands were reported.
C. Spatially Homogeneously Distributed Chain Scissions
The scission of molecular chains generally leads to highly reactive chain end
radicals and eventually to the formation of new end groups (cf. Chapters 6 and 7).
Unless oxygen is excluded the following end groups are formed most frequently:
methyl, aldehyde, ester, carbonyl, carboxyl, or vinyl groups. All of these groups
have characteristic IR absorption bands (Table 8.1). Based on this consideration Zhur-
kov and his collaborators quantitatively investigated the build-up of new end groups
in loaded polymer films [16-18, 22-26]. They employed a double beam technique
the principle of which is illustrated in Figure 8.7. The optical densities D = Qn 10/1 of
a stressed (a) and an unstressed reference film (r) are compared:
183
(8.4)
Here k is the absorption coefficient, d the thickness of the film, and C the con-
centration of end-groups. Figure 8.8 shows such a difference spectrum recorded by
Zhurkov et a1. [17, 18] from a polyethylene ftlm ruptured at room temperature.
Table 8.1. IR Absorption Bands Suitable for End-Group Analysis [17,18,24,39].
(End) Group Structure Absorption Band Absorption Coefficient [241

(em-I)
II k (10- 19 em 2)
Vinyl -CH 2 -CH=CH 2 910 2.52

Vinylene -CH 2 -CH=CH-CH2- 965 1.46
Aldehyde -CH 2 -CHO 1070 2.52
Methyl - CH2-CH3 1379 1.02
Vinyl -CH2-CH=CH2 1645 0.68
Carboxyl -CH 2 -COOH 1710 8.92
Aldehyde -CH 2 -CHO 1735 2.47
Ester -CH 2-COO-CH 2 - 1745 6.36
Perkin - Elmer Spectrophotometer 221
r - - I - - - - - - - IR
I
I
I
t I
J
0°
Detector . .11_
o e=--=:-u=
./ I t
Polarizer' I tForce p
L___ , __ --- IR
l---------------Fllm
L -_ _ _ _ _ _ _ _----' * -------------Aluminium
Stretching DirectIOn
Fig. 8.7. Principle of double-beam IR technique.
The spectra have been evaluated with respect to the kinetics of end-group accumula-
tion and to the absolute number of chain scission events. The results will be reported
although it seems as if a full understanding has not yet been obtained in terms of
sample morphology, molecular weight changes and ESR results. So far only three
polymer samples (PE-LO, PE-HO and PP investigated in the Joffe-Institute) have
yielded end-group concentrations sufficient for detection. Thus the initial remarks
concerning comparison and transfer of results apply here.
184
0
N :r:
IU '" u
:r:"
:r:
u M
:r: '"
1740
u "
:r: ';' :r:
0
0
N
:r:
U
, N
:r:
0
u,
0 u
ri:
u
,
~if) OJO '"
"
0 1710
<J
910
AA
965 1379
920 960
A
1360
I:
1400
f-I ~ _ _ _
1720 v
~---l
1760 em'
Fig. 8.8. Difference IR spectrum of unstressed and stressed oriented polyethylene (after [16-18]).
f
-r----/f-----
- , - 3: .--·-r-4---
7 - zl'2- ·70-
I,.
.
0 .... 0
- o~o-=-
,
- -b
- -O-C
05-0~o __ ~_. _ _ _ _ a
",,0 i '
c O~--------+_----------~----~
o
.~
.g
C
1
b
8 O~ ______________________ ~ ____ ~
o
time
Fig. 8.9. Accumulation of aldehyde-end groups formed in high density polyethylene (after [16 -181).
(a) T = 20°C; a(MN/m2): 1. 360.2.350,3.330,4.280; (b) a = 280 MN/m 2 ; T (OC): 1. 67, 2. 46,
3.20.
Subjecting uniaxially oriented films to uniaxial loads Zhurkov et al. [16-18,

22-26] measured by such a technique the accumulation of aldehyde groups in high
pressure and low pressure polyethylene and in polypropylene after predetermined
loading intervals. Typical results are reproduced in Figure 8.9. The concentration
C(t) of aldehyde-end groups increased with loading time according to a first-order
kinetic equation:
C(t) = C* (l-exp-Kt) (8.5)
with rate constant K and apparent concentration C* of mechanically excitable -

and thus breakable - bonds. The authors found that the concentration C* depended
only slightly on temperature and stress (changes from 1 to 4 . 10 19 cm -3 for stres-
ses between 180 and 500 MN/m 2 at temperatures between -70°C and +20 DC)
whereas the rate "constants" K varied by several orders of magnitude (Fig. 8.10).
The authors noted that the rate constant K for accumulation of end groups was
governed by an effective, stress-dependent activation energy E* which is plotted in
Figure 8.11. Using the apparent linear relation between E* and a one obtains:
185
K = Ko exp [-(EA - Q* a)/RT]. (8.6)
The activation energies EA and activation volumes Q* for the three polymer
films investigated are listed in Table 8.2 together with the preexponential factors K o.
In these experiments the authors also observed that the concentration of end
groups at break for a series of specimens did not depend on stress, temperature, or
type of loading (Fig. 8.12). The apparent "critical concentration" of all newly formed
end groups inPE-HD amounted to 3.9·10+19 cm- 3 or to ~ 2·10+19 bond ruptures
-4
-8
-12
~---r---+----+----r~~
12 4 !Q' K-1
T'
a
Fig. 8.10. Rate constants K for end-group accumula-
-4 tion and times to fracture T for high density poly-
ethylene (after [16-18]); a(MN/m): 1. 400,2.300,
3.200,4.400,5.300,6.200.
---
keal ,
moL
-0-
U*
E* I
30 ------
--
I
-- --
>.
~.,
c.,
c 20 ·-~-.......o--'-" I
----
0 I - __0
. r-'-'-J
~>
~
'" 10
a I I
a 200 400 G N/mm2
stress
Fig. 8.11. Apparent activation energy. U. for macroscopic breakage under uniaxial stress a;
E. for accumulation of end groups under stress a (after (181).
186
Table 8.2. Kinetic Parameters of End Group Formation in Certain Polymer Films
(after Zhurkov, [18]).
Polymer Preexponential Activation Activation

factor Ko (s-1) energy EA volume Oi
(kcal/mol) kcal mm 2 lO-6 m3
---
Nmol mol
Polyethylene 1011 27 0.026 110

(low-pressure)
Polyethylene 10 12 28 0.059 250
(high-pressure)
Polypropylene" 10 12 30 0.051 210
Cb.1O-1~ em' 3
6 ,-- ---- - - - - -
0).., c
o o
o
-"
0
f; e 4 ~-~(f~=.--'O;;F'.---== +~ ~=®--
"""<V <V
C o 0 0
~
.0 5 2 1--- - - t - - - - t - - - - - - - -- -
"
o I I
~O----~20~O------~3~OO~-----4~OO~~ r-6--~N~/-m-m~2
stress
Fig. 8.12. Concentration Cb of end groups just prior to macroscopic fracture of high density
polyethylene (after (181); conditions of loading: samples broken in creep at constant stress at
(+) 200 K; (e) 273 K; (0) 293 K; (D) 373 K; (e) 333 K; (A) 323 K;
(~) samples broken under linearly increasing deformation with time; (Ell) samples broken under
stress increasing linearly with time.
per cm 3 . Vettegren and Chmel [70] estimated that in a surface layer of 1 J1111 thick-
ness the concentration of end groups may even be higher by a factor of 20 to 60.
For polypropylene they derived concentrations of 6· 10 19 cm- 3 broken bonds in
the surface layer - as compared to 0.25 - 0.40· 10 19 cm- 3 reported by Zhurkov
et al. [17, 18] for the interior of the film. The latter value should be compared with
the number of overloaded 5 nm segments. In Section I B a concentration of
8.5 . 10 19 cm- 3 overloaded segments had been determined from the area of the
tail region of the "stressed IR band". Such a comparison indicates that of
No = 56· 10 19 cm- 3 segments present in the interior material of the PP film 0.15 No
are initially overloaded and 0.0045 No will have been broken at the point of mac-
roscopic fracture. In another study on PP [25] a concentration of broken bonds
of 3· 10 19 cm- 3 is reported, i.e. a fraction of 35% of the overloaded bonds is broken
throughout the total volume of the sample. In the latter case wholesale destruction
of the polymer must have occurred since on the average each molecule in a (mono-
disperse) Mn = 50000 g/mol sample would have been broken 2.5 times. In the for-
mer case the effect is ten times smaller, but Mn is still reduced to 40000 g/mol and
Mw to 43750 g/mol. The effect is even more drastic with higher initial molecular
weights. The fracture criterion derived at the Leningrad Institute for the three poly-
187
mer fIlms on the basis of the kinetics of thermal and mechanical destruction and of
end-group formation is that of a critical concentration of newly broken chains
[16-18].
In view of the fact that so far the end-group analysis has only been successful
with these three polymer fl1ms one would welcome further studies on the extent to
which the end-group formation depends on the initial concentration of carbonyl
groups - or of other oxygen-containing groups. More extensive molecular weight
measurements, especially by fractionating methods would also be useful to corro-
borate the absolute concentrations of end groups [220].
If the number of breaking chain segments is as large as indicated by the end-
group analysis and if each segment breaks at the limiting stress value derived from
the analysis ofdeformed-IR bands then the accumulated molecular stresses would
be comparable within an order of magnitude to the applied macroscopic stress.
In that case it must be assumed that apart from conformational rearrangements and
chain slippage chain scission events noticeably influence observed stress-strain curves.
So far the PP material from Leningrad is the only polymer which has lent itself suc-
cessfully to both of the above mentioned IR analysis. In the cited literature [4-33]
there is no reference to an attempt to explain stress-strain or stress-time data of this
PP material on the basis of its end group-formation kinetics.
As discussed frequently within this text stress-relaxation data can not and must
not be related solely to chain scission [2-52]. Attempts have been forwarded, how-
ever, to explain the stress-strain curves of polyamide-6 [49-51] and PEOB fibers
[5 2] on the basis of free radical production kinetics. In these models the concept
was employed that tie segments show a distribution of relative lengths. The (taut-
ness) distribution, N(L/Lo), was assumed to remain unchanged in a wide time-tem-
perature regime. As detailed in Chapter 5, a tie-chain segment will respond to a strain
€a predominantly rubber-elastically if €a < (L-Lo)/Lo, it will break if €a > (1 + 1/Ib/
Ek)L/Lo - 1, and it will respond energy-elastically in the intermediate region where
it happens to be fully extended and sub critically loaded. From these considerations
a homogeneous four-phase model can be derived (Fig. 8.13). It is based on the assump-
Fig. 8.13. Four-phase model of highly oriented semi-

crystalline fiber. (a) crystalline lamellae, (b) fully
extended tie segments, (c) non-extended tie segments,
d (d) matrix (folds, ciliae, broken chains).
188
I. Small-Strain Deformation and Fracture or Highly Oriented Polymers
tion that a highly oriented semi-crystalline fiber can be represented by crystalline

and amorphous regions in series with the latter consisting of three fractions, i.e. of
the fully extended segments, of the non-extended tie-chains, and of the remaining
matrix (folds, ciliae, broken chains). Once the strain-dependent widths of the frac-
tions are obtained from a relative segment length distribution and appropriate moduli
are assigned to the individual phases a stress-strain curve can be constructed. Such a
procedure was proposed in 1971 by DeVries and Williams [49]. It has been executed
by Lloyd [50] and by Klinkenberg [51] for PA 6 fibers and by Nagamura et al. [52]
for PEOB multi-filaments. The calculated stress-strain curves can be made to fit the
stress-strain response of virgin material; they contain as possible sources of uncer-
tainty the choice of the "elastic" moduli and of the number and distribution of tie-
segment lengths, i.e. of N (L/Lo). The choice of the form of N (L/Lo) determines
which number of segment lengths is assigned to the fraction of non-extended chain
segments by way of extrapolation. Lloyd [50] and Nagamura et al. [52] have chosen
a normal distribution to represent N(L/Lo).
A fit of observed and calculated stress-strain curves of first and second stretching
cycles was obtained if the calculated number of radicals formed (or of chains broken)
was taken as considerably larger than the observed number. Nagamura et al. [52]
report a factor fc of 40 for PEOB fibers, Klinkenberg [51] of 20 for PA 6 fibers.
If one wishes to explain such a discrepancy while maintaining the basic model
one is forced to make additional assumptions:
- either the number of broken chains is systematically larger by fc than the number
of observed radicals, e.g. due to a Zakrewskii-mechanism [20] to be discussed
later;
or the breakage of N 1 chains in a particular volume element leads to the preferen-
tial unloading of fc N 1 extended chains; this amounts to the consideration of a
microfibrillar substructure;
in any event provision must be made to account for the anelastic deformation
of the non-extended chains; the stress-induced change of the conformation of a
tie-segment profoundly reduces the axial chain stress while not effecting N(L/Lo)
(cf. Chapter 5 IB and III); in subjecting a fiber to several stress-strain cycles a
large fraction of the more extended relaxed, conformations formed during the
first cycle are present at the beginning of the second cycle; these stress-induced,
temporarily stable conformational changes are considered to be mostly responsible
for the differences between first and second loading cycles [2].
It is the common observation of the above authors [49-52] that qualitatively
the slopes of stress-strain and of radical concentration-strain curves correspond. For
a quantitative correspondance one would have to assume that 20 to 40 times as many
chains break as free radicals are observed. Such an assumption appears to be rather
large in view of the facts that no correspondingly strong decrease of the molecular
weight has been reported and that the load bearing capability of the fiber material
outside of the immediate fracture zone does not necessarily suffer. Prevorsek [53]
showed that the strengths of fiber segments subjected repeatedly to breaking loads did
not decrease. The rupture segments obtained in a first loading exhibited a larger
strength than the original fiber and maintained this strength in subsequent loadings
(Fig. 8.14). It seems more likely, therefore, that the concentration of chain scission
189
,........, U8
• •
:-E •
I> I> I>
z .~
t!)
'--' 0.6
inititlal strength
.<::
0, I>~
c:
'" 0.4
!:
en
0.2 r-----.
.. repeated breakage"
o I
o 2 4 6 8
number of experiment
Fig. 8.14. Load-to-break of PA-6 fiber segments strained repeatedly (after [53 J).
events is of the order indicated by the ESR investigations, i.e. 10 16 to 5· 10 17 cm- 3 .

Such concentrations in themselves do not weaken the polymer network decisively since
they amount to the breakage of only 0.002 to 0.10% of all amorphous chain seg-
ments [2]. Unless the qualitative agreement between the slopes of stress-strain and
free radical-strain curves is coincidental it can be sought in an amplification effect
accompanying the rapid release of stored elastic energy. The ensuing local temperature
rise will facilitate the extension of kinked chains under annihilation of kinks which
is practically equivalent to unloading those chains without breaking them (cf. Chap-
ter 5). Within the framework of this model one would have to conclude that the
breakage of one amorphous segment at maximum load ql/lb leads to conformational
changes in the surrounding segments of the same microfibrillar region resulting in
local load decreases of 40 q I/Ib' In view of the fact that on the average there are
more than 2000 surrounding segments (of 5 nm length in any amorphous region
400 nm 2 in cross-section) the value of 40 q I/Ib appears very reasonable.
One word of caution must be said with respect to the extrapolation of the seg-
ment length distribution N(L/Lo) into the unknown regions of L/Lo . For this pur-
pose DeVries et al. [49, 50] and Nagamura et al. [51] used the assumption that the
"visible" part of the segment length distribution (Fig. 7.16) constitutes the skewed
part of a normal distribution. A plot of the skewed distribution on probability
coordinates (Fig. 7.17) apparently does not disprove the contention made. They had
derived the parameters of the normal distribution (mean, variance) using the existence
and location of an inflection point in the - always only partially known - cumula-
tive distribution. It cannot be excluded that such inflection points result from the
increased unloading of microfibril ends at larger strains since the stress-transfer length
Le of a micro-fibril (with modulus Ef , diameter d) increases in case of perfect inter-
fibrillar adhesion (of strength 7) with Le = Ef d f €/27. The eventual error of an extra-
polation of N (L/Lo) into the range of not fully - or even slightly - loaded segments
is of not too much importance in calculating accumulated molecular stresses. It is
of importance, however, if the distribution N(L/Lo) is used to derive a fracture
190
criterion. A fracture criterion in which macroscopic failure coincides with breakage

of the last tie segment of the total distribution must be ruled out. Against such a
fracture criterion the above mentioned argument holds that at the point of macro-
scopic failure 99.7% of all tie-segments are still unbroken (Fig. 7.6).
Although DeVries, Lloyd, and Williams [49, 50] work with a normal distribu-
tion of relative chain lengths they do not put it to a full and questionable use. Their
assumptions may be reformulated on the basis of the above considerations. It is
assumed 1. that the segment length distribution N(L/Lo) - derived from spatially
homogeneous chain scission events of virgin material - be representative for the
homogeneity of molecular stress distribution in any average amorphous region of
the fiber, 2. that an initially narrow distribution continues to rise sharply with seg-
ment length and 3. that the complete break-down of tie segments within micro-
fibrillar amorphous regions be restricted to a comparatively small fracture zone
which does not contribute measurably to the free radical population. They use the
widths of their chain length distributions N(L/Lo) and correlate it with macroscopic
strength (Fig. 8.15). In other words they correlate the non-homogeneity of the
molecular stress distribution with macroscopic strength - and find a negative cor-
relation.
MN
°I------- nylon fiber
results
I
m2
\
'00",,-
°0
~1000
iii 900
E
E800 ~,
~ 700 0"'J,
600 ..........
SOO ..........................
400 nylon '0 56"

300 t)(ncilrg matenal
10 1.20 1.40 160 180 2 00 240 2.60 2.80 300 3.20 340
one standard devlallOn value In Of. stram
Fig. 8.15. Correlation of the homogeneity of segment lengths (narrow relative length distribution)
with uniaxial strength Gb [49,50,541.
It has been stated frequently in this work that the tie segments which eventually
break are not the main source of strength of highly oriented fibers. The total number
of radicals formed at macroscopic fracture does not measure, therefore, the strength
of the sample. The wide differences in radical concentrations N(R) at break of com-
parably strong fibers (cf. Table 6.2) illustrate this statement. Again these concentra-
tions should be taken only as an indication of the homogeneity of stress distribution
between different rnicrofibrils and of stress response within different micro fibrils.
The later a possible defect zone starts its accelerated growth the longer the phase of
spatially homogeneously distributed chain scission and the larger the number of
191
chains which are loaded up to breakage. If one compares, under this aspect, members
from one family of samples strength-related conclusions seem to be justified.
In Chapter 7 the morphological changes in PA 6 fibers due to annealing and
their effect on chain scission have been discussed. In Figure 7.18 and 7.19 it had
been indicated that slack annealing is accompanied by a relative lengthening of tie-
segments and to a widening of the length distribution. The indicated losses of homo-
geneity lead to accelerated flaw growth already at stresses lower than those exerted
on the control sample, i.e. to a loss of strength (Fig. 7.20, "slack"). Annealing at
fixed fiber ends leads to a certain loss in homogeneity while maintaining the average
relative segment length. The loss in strength (Fig. 7.21) and in the number N of radicals
at break (Fig. 7.20) as compared to the control sample is small. This result confIrms
the findings of Levin et al. [21] with highly oriented tensile Kapron annealed at 200 0 C.
Annealing under all. 7% prestrain increases the relative segment length (Figs. 7.18
and 7.19), segment homogeneity (Fig. 7.22), and sample strength (Fig. 7.21).
The opposing effects of increasing annealing temperature and increasing tensile
prestrain on segment homogeneity are clearly indicated on Figure 7.22. The concen-
tration Nit of radicals at break formed after annealing initially increases with pre-
r
free radicals - - - - I '
at prestrain
I
/
/'
0.7 '--_:'::-_ _-'--_----'_ _--"'.::..0=--_:'----1 0
-10 0 10
1IL1L[%]
Fig. 8.16. Effect of prestrain at annealing at 164°C on strength and number of broken bonds of
6 polyamide fibers (after [55-571).
192
strain (Fig. 8.16), showing a maximum at a 5% prestrain. At that strain radical forma-
tion in prestraining (Np ) starts. The total number of radicals (Np + NR) is approxi-
mately constant in that region. The sample strength increases by 20% up to a maxi-
mum at 7.5% prestrain and decreases slightly thereafter (Fig. 8.16).
D. Formation of Microcracks
All of the foregoing considerations have been concerned with a distribution of chain
scission events homogeneous on a macroscopic scale. The accelerating effects of
stress concentrations at microstructural inhomogeneities and of mutual interaction
of scission points have so far been neglected. Such effects were not apparent in the
discussed IR and ESR investigations of chain rupture; in any event they did not
prevent tremendous concentrations of scission points being accumulated before
final fracture. This fact can certainly be taken as an indication that during a con-
siderable period of straining, chain scission points remain isolated "defects" and
do not give rise to unstable cracks.
On the other hand the appearance of submicrocracks in loaded polymers has
been known for many years since the Leningrad school [17, 18, 27, 28] employed
X-ray scattering techniques to search for it. Such submicrocracks were found in PE,
PP, PVC, PVB, PMMA, and PA 6. The authors noted two important regularities of
the sub micro crack formation [28). First of all, the submicroscopic cracks appear at
finite sizes with their transverse dimensions practically independent of load duration,
stress value, and temperature (Table 8.3). Secondly the transverse size of the sub-
microcracks is determined by the polymer structure. For oriented crystalline poly-
mers the transverse size coincides with the microfibril diameter; for non-oriented,
amorphous polymers having a globular structure it coincides with the diameter of
the globules [28].
Two explanations have been given for the formation of these submicrocracks:
Zakrewskii [20] proposes a radical chain reaction and Peterlin [58] suggests that the
ends of micro fibrils are the crack nuclei.
Table 8.3. Characteristics of Incipient Cracks in Polymers (after Zhurkov Kuksenko

et aI., [27,29])
Polymer Micro crack diameter Ncr Density change IIp/ p

longitudinal transverse cm- 3 calculated measured
nm nm
Polyethylene 15 17 6.10 15
Polypropylene 20 32-35 7· 10 14
Polyvinyl chloride 60 8· 10 14
Polyvinyl butyral 50 3.10 14
Polymethylmethactylate 80 170 4.10 12 0.4.10- 2 0.3.10- 2
Polycaproamide 5 9-25 5 . 10 16 1,8.10- 2 2.0.10- 2
193
The Zakrewskii-mechanism was proposed by the Russian school following their

studies of the formation of free radicals and of end-groups. They had noted that
the concentration of free radicals was either by a factor of 4000 (PE-HD) smaller
than the concentration of newly formed end-groups or even unmeasurably small
(PP). This led them to search for a reaction by which chains might break without
increasing the concentration of free radicals.
The mechanism of the proposed mechano-chemical reaction chain is illustrated
in Figure 8.17. The thermo-mechanical scission of a regular covalent bond (a) leads
to the formation of two chain-end radicals (b). As amply discussed in Chapter 6 the
latter radicals are highly reactive. They will transform by hydrogen abstraction into
stable end groups at the same time creating main-chain radicals within neighboring
chains (c). It is then assumed by Zakrevskii and Zhurkov [17, 20] that the decreased
energy of activation for the thermal scission reaction of radicalized chains (cf. Chap-
ter 4) also leads to a lower tensile strength of these chains causing them to rupture
(d). Repetition of the steps c and d would not increase the number of free radicals
present but would mUltiply the number of chains broken. In Figure 8.18 the energy
levels of the proposed steps of the reaction chain are indicated. It should be noted,
however, that the lowering of the activation energy for scission of a radicalized
chain is effective only where thermal eqUilibrium can be established (cf. Chapter
4 II B).
a b
Fig. 8.17. Mechano-chemical reaction chain after Zakrevskii and Zhurkov [17,20,551: (a) stress-
induced chain scission, (b) formation of chain-end radicals, (c) radical reaction leading to main-
chain radicals, (d) scission of radicalized chains, (e) formation of a submicrocrack by repetition
of the steps (c) and (d); III chain-end radical e.g. -CH 2-CH 2, x main-chain radical e.g.
-CH2-CH-CH2-, • stable end groups e.g. -CH 2-CH 3.
The authors suggest [17,18] that the 3300 new end groups formed per sub-
micro crack in PE-HD (5000 in PP) are created by such a mechanism. Their data in-
dicate also that there is less than a single pair of free radicals available per submicro-
crack (0.4 in HDPE, 0.3 in PA 6 and an immeasurable small fraction in PP). These
discrepancies, the unfavorable conditions for thermal equilibrium in the splitting
reaction, and the absence of a well defined termination of the reaction chain raise
considerable concern. They make it difficult to believe that such a reaction is the
major reason for the formation of submicrocracks which remain fairly stable during
extended and repeated loading periods [220].
Peterlin [58] has based his proposal on the origin of sub microcracks on mor-
phological considerations and analysis of the previously mentioned X-ray data
194
kcol / mol
"0
co
.8 '"
:E ~
100 ..,o"
co
.9
"
2
E
"0 -c H2 -CHZ + CH2 -CHZ _ H -obstroctoon

>-
Q.
"5 50
""~
Fig. 8.18. Scheme of energy levels of steps a-d in the mechano-chemical reaction chain.
[17 -21,27]. He suggests that the ends of microfibrils, preferentially situated on the
outer surface of the fibrils, retract under stress (Fig. 8.l9). In low strength PA 6 they
thus open up about 10 16 oblate spheroids per cm 3 having 6 nm diameter in fiber
axis direction and 10 nm in the perpendicular direction. He estimates that the num-
ber and size of cracks in high strength polyamide is considerably smaller and is in
agreement with the much higher draw ratio employed in producing those fibers.
For highly oriented films with microfibrillar substructure similar considerations on
the origin of the voids hold.
Fig. 8.19 . Fibrillar model of fibrous structure with practically all ends
of microfibrils concentrated on the outer boundary of fibrils (after (581).
195
Considering the two important regularities of submicrocrack formation [28],

the inconsistencies of the Zakrewskii-mechanism [17], and the morphological as-
pects [58] the retraction of microfibril ends seems to be a logical interpretation of
the opening-up of fixed-geometry voids. This means at the same time that the sub-
micro crack formation is a process inherently independent of chain scission or end-
group formation [220]. The voids constitute structural irregularities and contribute as
such toward the general non-homogeneity of the distribution of stresses or strains.
Their direct influence as potential stress concentrators is weak and ineffective with
regard to accelerating chain scission. This conclusion is based on the following facts:
microfibril ends are free to undergo shear deformation and are free from axial
stresses,
obviously the voids do not prevent the build-up of large elastic strains in the fibers
or films - as evidenced by the scissioning of chains and the appearance of de-
formed-IR bands,
if chain scission would occur preferably adjacent to microfibril ends further lateral
widening of the there existing microcracks would then invariably accelerate chain
scission.
E. Energy Release in Chain Scission
A molecular chain segment stressed almost to its load-bearing capacity constitutes

an extremely powerful source of elastically stored energy. In thermo-mechanically
activated bond scission only a minor fraction of this stored energy is needed to break
the chemical bond, namely the mechanical contribution {31/ib to the activation energy
Uo. The remaining, major part of the energy is available for mechanical interaction
with surrounding chains or is diSSipated as heat. The diSSipated heat has a twofold
effect through the ensuing local temperature rise: it increases the mobility of other
chain segments and decreases their rupture strength 1/ib (T). Both effects tend to
facilitate the further degradation of the stressed polymer.
The total elastic energy stored in a segment of elastic modulus Eb length L,
and cross-section q amounts to
(8.7)
For a polyamide 6 segment of5 nmlength with q = 0.189 nm 2 , 1/ib = 21 GN/m 2

and Ek = 200 GN/m 2 one obtains a value of 1 .10- 18 J per segment or 600 kJ/mol
for the elastically stored energy. To break a C-C bond at room temperature a me-
chanical energy of 110 kJ Imol is needed (cf. Chapter 7 I C l). In setting up an energy
balance the contribution of the elastic forces must not be forgotten which hold the
highly stressed tie-segment ends within the crystal lamellae. At the rupture stress this
energy amounts to about 190 kJ/mol for each segment end (cf. Table 5.5). One thus
arrives at an energy of 870 kJ/mol which is liberated at the moment of chain scission.
If this energy were to be confined within the volume of the segment and of its ends
(Ltotal ~ 10 nm) it would constitute an energy density of Wtotadqtotal = 764 MJ/m 3 .
196
Chain scission events, therefore, have been justly termed micro-explosion by Zhurkov
and his colleagues.
On the other hand at a (large) concentration of (0.6 . 10+ 24) bond ruptures per
m 3 (Le. of 0.83 mOl/m 3 ) the total energy having been stored in the retracting chain
ends amounts to an average of 722 kJ per m 3 which will be dissipated as heat Qb.
The chemical energy Vb of this number of broken bonds amounts to 156 kJ/m 3 .
These average energy values should also be compared against the density of elastically
stored energy, i.e. against a 2 /2 E. This term reaches about 125 MJ /m 3 for very high
strength PA 6 fibers, which in tum is only one sixth of the cohesive energy density
of that material. The energy dissipation due to ruptured bonds is slightly smaller,
therefore, than the pure hysteresis losses encountered in loading and unloading a
polyamide fiber (for PA-6 at 10 to 30 Hz tan 8 is of the order of 5.10- 3 ).
The important problem of how the external work W done on a polymer sample
during a loading-unloading cycle serves to increase the internal energy V, to change
the entropy S, or to be irreversibly dissipated as heat Qir = TdiS was attacked by
MUller in 1956 in his classical experiments [59-62]. According to Eqs. 5.2 and 5.3
one can write
dW= dV - TdS - Qir. (8.8)
Miiller and his co-workers measured several deformation-related thermodynamic

quantities: the reversible and irreversible parts of heat production during the plastic
deformation ofPE, PVC, PETP, PA 6 [59-61], PC [63], PS [64], and of various
elastomers [61, 65, 66], the ensuing temperature rises [67], the change of internal
energy with the time of storage [68] and its effect on the energy of dissolution [69].
They noted that the entropy of thermoplastics during cold flow is decreased and
that the internal energy is increased. They also measured the balance of energy in
stressing (s) and during retraction (r) of PIB in a sequence of stretching cycles. The
change of internal energy in an i th cycle can be written as:
(8.9)
of which the entropy terms are equal by definition. The authors observed that 8 Vi
was different from zero only in the first cycle. The increase of internal energy during
extension of elastomers and thermoplastics was ascribed by them to a reduction of
close range order and to chain scission and formation of radicals as further sinks of
external energy.
The latter hypothesis was further tested by Godovskii et al. [31]. PA-6 fibers
drawn to A = 5.5 at 210°C were repeatedly stretched at room-temperature. The
authors found for PA 6 the same characteristic difference between the first and the
following loading cycles as did MUller for PIB: 8 V was essentially different from
zero only in the first loading cycle. The ratio of 8 WI /8 VI was found, independent
of the macroscopic stress, to amount to 7.0. This constancy of 8 WI /8 Viis in fact
remarkable. It indicates that processes are responsible for the increase of internal
energy which occur, independent of a, if locally a critical excitation is surpassed.
Godovskii et al. assume that these processes are chain ruptures. They derive from
197
8 U1 the number N 1 of chain scission events each of which contributes an increment

of internal energy of 1.7· 10- 19 J (100 kJ/mol). The dissipated energy per chain
scission event, 8 W1 /N 1 , amounted to 700 kJ /mol. These figures are only slightly
smaller than the values found above on the basis of elasticity considerations for the
mechanical contribution to chain scission (110 kJ /mol) and for the energy dissipated
by the retracting segments (870 kJ/mol). This striking coincidence does not prove,
however, the above hypothesis that 8 U1 might be interpreted solely as an increase
in chemical bond energy due to chain scission.
Millier [62] and Kausch and Becht [55] have indicated that slip processes and
conformational changes may respond energetically similar to bond ruptures if they
are occurring as a consequence of stress induced distortion of the polymer network
or as a result of local heating. The increase in internal energy is then related to a
decrease of close range order or the decrease of the number of hydrogen bonds or
to internal stresses between chains and within crystallites. The existence of inter-
molecular forces of considerable magnitude in oriented polymers can be inferred
from a number of optical, spectroscopical and mechanical experiments. Particularly
noteworthy are the following observations. Vettegren et al. [70] noted that in per-
fectly annealed PETP film the maximum of the characteristic backbone vibration
band lies at 975 cm- 1 , whereas in oriented PETP film it lies at 972 cm- 1 . Lunn and
Yannas [71] found that the asymmetric stretch vibration band of methyl groups
with a maximum at 2971 cm- 1 in isotropic PC film splits into two components
1 cm- 1 apart in a PC film drawn to about 85%. They assigned this frequency shift
to the action of intermolecular rather than intramolecular forces. With the drawn
material they also studied chain-segment orientation by determining the dichroic
ratio of the 1364 cm- 1 band. They obtained an interesting result that chain back-
bone motion (with an ensuing change of the degree of orientation) in drawn PC films
was clearly observable more than 60 K below the glass transition temperature of
149°C. The largest deviations from an equilibrium state (highest intermolecular
tensions) existed in those films drawn at the lowest temperature (23 °C) were anneal-
ing at 81 ° C resulted in measurable changes of the orientation distribution. Millier
[62] determined that in the cold drawing of non-oriented polymers the fraction
8 U/8 W of internal energy rise to external work amounted to 0.2 (PE), 0.3 to 0.45
(PA 6, PETP) and 0.3 to 0.5 (PVC).
The question is still open, therefore, as to the extent the internal energy rise
is due to the chemical energy of broken bonds or to the local rearrangements trig-
gered by chain scission.
F. Fatigue Fracture of Fibers
The preceeding Sections have dealt exclusively with the response of chains
and micro fibrils to constant or monotonically increasing stress or strain. In service,
however, fibers are frequently subjected to intermittent, alternating, or cyclic loads.
The behavior of fibers, therefore, to repetitive cyclic loading has been studied for
many years (e.g. 72-82). Following an extensive review of Hearle et al. [76] it can
be said that in cumulative extension cycling a fiber fails by virtue of having reached
198
its breaking extension. Under such a condition of constantly increasing maximum

extension, fatigue can be predicted from an adequate knowledge of the anelastic
deformation and the time-dependent breakage conditions of the fiber. As yet no
specific fatigue effects have been identified in cumulative extension cycling [76].
In one of the earliest reports on fatigue of fibers and fabrics Busse et al. [72]
pointed out that the vibration life of nylon, cotton, and rayon cords is inversely
proportional to the rate of thermally activated flow steps.
A similar conclusion has been drawn by Regel et al. [74] and Tamush [75] for
cyclic stressing. These authors assume the validity of the principle of damage ac-
cumulation ("Miner's rule"), i.e. of
21TNF/W
f dt / T(a[t]) = 1 (8.10)
o
with NF the number of cycles to failure and w the loading frequency. If T, the expec-
tation value of the time-to-break under constant stress, is taken to be the common
Arrhenius expression one obtains
21TNF/W
f dt / TO exp [(Uo - 'Y a [t])/RT] = 1. (8.11 )
o
Regel et al. [74] report that such a damage accumulation law did apply to PAN
fibers loaded at 24 Hz up to 1.5 . 10 7 cycles. For PMM films, viscose fibers, and
capron fibers (PA 6) correspondance of experimental data with Eq. (8.11) could be
obtained through air cooling of the fatigued samples, after preliminary drawing, or
at elevated temperatures. The authors concluded that Eq. (8.11) would describe
fatigue fracture within the kinetic concept of fracture if (ambient) temperature T
and activation volume 'Y would be replaced by values T* and 'Y* which would depend
on the parameters of the fatigue experiment (frequency, form of stress or strain pulse)
The fatigue failure of fibers has also been extensively investigated by Prevorsek,
Lyons, and co-workers. Their many pertinent papers are referenced in [78]. The
authors interpret the fatigue life as the time to void nucleation by rearrangement of
molecular segments. They arrive at a kinetic equation relating the number of cycles
to failure with various mechanical and molecular parameters [78].
Employing a load-controlled fatigue tester Bunsell and Hearle [77] have more
recently found a distinctive fatigue fracture morphology in polyamide 66 fibers. A
necessary requisite for this fatigue mechanism to operate was the cycling of loads
down to zero stress values. Under such conditions fibers failed at a maximum load
only 60 to 70% of the normal breaking load [77, 79]. This observation will be further
discussed in the following section (8 I G).
DeVries et al. [49] studied the formation of free radicals under cyclic tensile
loading. The radical accumulation curves reflected the stress cycles. The authors
explained the kinetics of radical accumulation on the basis of Eq. (7.2), i.e. with-
out assuming a special fatigue mechanism.
Summarizing the above observations one has to conclude that no distinctive
fatigue mechanism - other than hysteresis heating - is operating when highly ori-
199
ented fibers are subjected to cyclic loads which remain always positive. Initiation
and propagation of fatigue cracks will be favored, however, if chains and fibrils are
given a chance to relax stresses, to degrade possible strain hardening effects, and to
reorient. Thus the fatigue mechanism described by Bunsell and Hearle [77,79] mani-
fests itself as an increase in interfibrillar slip and crack growth almost parallel to the
load direction. It will be further discussed in the following section. Characteristic
fatigue mechanisms are also well observable in non-oriented polymers. They will be
treated in Section II C of this and in the next chapter.
G. Fractography
The visual inspection and the study of broken samples by means of optical or scann-
ing electron microscopy (SEM) are important tools of fracture analysis. They ob-
viouslyaid
- in the detection of possible causes of crack initiation and
- in the interpretation of the fracture process.
The appearance of a fracture surface undoubtedly is the most convincing evidence
that the fracture process has reached the phase of non-homogeneous deformation.
But rather frequently the surface and morphology of a failed specimen reveal whether
or not a phase of homogeneous material deformation has contributed to the fracture
event. It is in this sense that a paragraph on fractography has been introduced into
the chapter on homogeneous deformation and fracture. Figures 7.8 and 7.9 serve
to illustrate this point. These fracture surfaces are more or less an arbitrary result
of a large number of homogeneously distributed scission and slip processes of chains
and microfibrils. The fracture surface was developed within a very small fraction of
the total lifetime and at a location not predictable in advance.
Hendus and Penzel [83] investigated the fracture morphology of polyamide 6
monomaments. The regularly spun and drawn monomaments were subsequently
subjected to tensile tests at various strain rates. Characteristic fracture surfaces are
reproduced in Figures 8.20 and 8.21 [84]. At small strain rates (e = 0.033 S-I) fre-
quently v-shaped notches are observed (Fig. 8.20). Such a notch is formed through
a crack which initiates at a flaw or material inhomogeneity contained in the ma-
ment surface or in a zone close to the surface. While the crack grows slowly the re-
maining fiber cross-section continues to deform plasticly. At a point determined by
the sizes of the crack and of the remaining cross-section and by the material pro-
perties rapid transverse crack propagation occurs. The measured strength of the mono-
mament is the higher the smaller the v-shaped notch [83]. Filaments of highest
strength contained barely visible small voids.
Figure 8.20 also shows an enlargement of the head of the monofilament. This
enlargement corresponds to a shrinkage of the oriented material caused by the warm-
ing of the mament during the plastic deformation. If the rate of deformation is in-
creased to 50 S-1 then the heat generated through the plastic deformation of the
remaining cross-section cannot be dissipated fast enough. Locally the melt tempe-
rature is exceeded, the mament heads expand almost to the diameter of the undrawn
material (Fig. 8.21).
200
l. Small-Strain Deformation and Fracture of Highly Oriented Polymers
Fig. 8.20. Corresponding ends of PA 6 monofil broken at a strain rate of 0.033 s-1 (Courtesy
H. Hendus, (831).
Fig. 8.21. Corresponding ends of a PA 6 monofil broken at a strain rate of 50 s-1 (Courtesy
H. Hendus, (831).
The fibrillar nature of drawn monofilaments becomes apparent in Figure 8.22.

The filament ends of two different fracture events show a strong axial splitting. As
concluded from the formation of small heads at the ends of some microfibrils the
fibrillation must have occurred before the catastrophic fracture took place. Strong
axial splitting is also known from polyamide 66, Kevlar [148], PETP, acrylic fibers,
wool, human hair, and cotton fibers [85] .
On the basis of fractographic studies Bunsell and Hearle [77, 79] indirectly
identified a fatigue mechanism in polyamide 66. Oriented PA-66 fibers failed in a
tensile type manner if they were fatigued uniaxially between an upper stress level
equal to the long-time static strength and a lower level of clearly non-zero tensile
stresses. These fatigue failures had a morphology similar to that shown in Figure 8.20.
The breakage apparently did not involve any specific fatigue effect. If the loading
conditions were changed, however, so as to include zero tensile or even compressive
201
Fig. 8.22. Two different PA-6 fiber fractures under tension showing strong axial splitting
(Courtesy H. Hendus, [83}).
stresses a specific fatigue fracture morphology became apparent. Those breaks, oc-
curring after 105 cycles, were characterized by the propagation of a crack, almost
parallel to the direction of the fiber axis. As shown by Figure 8.23a [86] this leaves
a long tail of material at one fiber end. Although this phenomenon is not yet com-
pletely understood it may be concluded from Bunsell's and Reade's observations
[77, 79, 85] that compressive yielding and/or buckling of microfibrils or fibrils takes
place on a microscopic scale. The repeated lateral rearrangement leads to a gradual
loss of interfibrillar cohesion. A crack which has started at some flaw in a direction
normal to the tensile stress a will turn, therefore, into a direction u almost parallel
to a - the angle between u and a being a function of the lateral stress transfer
which still remains.
Fig. 8.23 a and b. Corresponding ends of fatigue failure of polyamide 66 under oscillating load
dropping to zero in each cycle (Courtesy J. w. S. Hearle, Manchester).
202
1. Small-Strain Deformation and Fracture of Highly Oriented Polymers
On the basis of his extensive investigations Hearle [85] has established a classi-
fication of the main features of fiber fracture morphologies:
A. Transverse Elastic Crack Propagation
This is the classic form of failure in elastic materials and consists of a relatively
smooth "mirror zone" of crack propagation leading to a rougher zone of final
failure due to multiple crack-initiation (comparable in morphology to Fig. 1.7).
B. Ductile Transverse Crack Propagation
This is a form of fracture in which a crack propagates stably across a fiber under
increasing load and/or strain and is opened out into a V-notch by the continued
plastic yielding (the final stages of the drawing process) of the remaining material;
the crack leads into a region of final catastrophic failure, occurring when the stress
in the remaining reduced cross-section reaches a critical level (Fig. 8.20).
C. Transverse "Fibrous" Break
This type of break runs perpendicularly across the fiber, with a rough texture
and no evidence of any propagating cracks; the whole structure appears to be
ready to fail at the same time, and the breaks are very similar in appearance to
lower-magnification views of the fracture of fiber-reinforced composites (Fig. 7.8
and 7.9).
D. Axial Splitting under Tension
In some fibers failure in axial tension occurs by cracking, or splitting, along planes
close to the fiber axis. Due to a nonhomogeneous fibrillar substructure there will
be shear stresses in these planes. With reduced interfibrillar cohesion a break is
produced characterized by multiple axial splitting over a long length (equal to
many fiber diameters) (Fig. 8.22).
E. Splitting Due to Torsion
Under shear stresses generated by torsion, splitting can occur along lines deter-
mined by the directions of stress and of material weakness.
F. Axial-fatigue Cracks
Particular localized cracks, deviating slightly from the fiber-axial direction,
develop in many fibers as a result of tensile fatigue: they are associated with shear
stresses at discontinuities. The resulting breaks show a long tail on one end, a strip
off the other end, and a final catastrophic-failure region (Fig. 8.23).
G. Fatigue-cracking along Kink-bands
Kink-bands at angles of about 45° develop in many fibers in compression, for
example, or on the inside of bends, and, in repeated flexing, these eventually turn
into cracks and lead to fiber failure.
H. Multiple Splitting in Fatigue
In torsional- or flexural-fatigue situations, multiple splitting frequently occurs
and can eventually lead to failure.
For a more detailed discussion of fiber fracture morphology the reader is referred
to the series of publications by Hearle, especially to "An Atlas of Fibre Fracture"
[87]. The fracture morphology of non-oriented polymers under various loading con-
ditions will be treated, where appropriate, in subsequent sections of this and the final
chapter.
203
II. Deformation, Creep, and Fatigue of Unoriented Polymers

A. Impact Loading
Polymeric materials are viscoelastic solids. Their propensity to anelastic and plastic
deformation is reduced when they are tested at high load rates and/or at low tem-
peratures. The reduced deformability causes a formerly tough or highly elastic poly-
mer to respond in a brittle manner. The impact fracture of natural rubber at liquid
nitrogen temperature is convincing evidence of this fact.
The resistance of polymeric materials to impact loading is of considerable tech-
nical importance as illustrated by the following examples:
aircraft windows have to withstand impinging particles or rain
protective coverings of high voltage switches must resist the impact of metal chips
water pipes or plastic floor tiles should not be damaged if hit accidentally
car bumpers or packaging materials must absorb as much of the energy of a col-
lision as possible
- household items are not expected to break the first time they fall to the ground
gears should transmit even rapidly increasing forces
- safety belts should yield but not fracture.
In view of the technical relevance the impact resistance of polymers has been
intenSively investigated. The cited references [88-103] may serve as an introduction
into the large body of literature. Vincent [88] and Bucknall et al. [89] give a general
survey of the impact testing of polymers. Other references concern the molecular
aspects [88-96], instrumentation [97-100], and particle impact [101-103]. The
extensive literature on the fracture mechanical analysis of impact bending tests will
be referenced in Chapter 9.
The fact that the impact resistance depends strongly on the presence and shape
of stress concentrators, on the sample geometry, and on the testing conditions has
made it very difficult to define and measure a unique material property impact
resistance. Since the impact behavior under one condition (e.g. impact bending)
cannot very well be predicted from the behavior under different conditions (e.g.
falling weight) quite a number of impact tests have been devised. The four best known
ones are the three point bending (Charpy unnotched and notched), two point bend-
ing (Izod), tensile-impact, and falling-weight tests which have been standardized
(DIN 53453, 53448,53373, 53443E; ASTM D 256, 1822,2444,3029).
In the context of this chapter-only the three-point bending test (Charpy) will be
discussed. In this test the energy loss An is measured which a pendulum incurs in
striking and breaking a prismatic specimen (of thickness D and width B). The ob-
served energy loss comprises basically four different terms:
the energy We to bend the sample up to the point of crack initiation,
the energy GcBD to propagate the crack through the specimen,
the kinetic energy Wkin of the broken sample, and
vibrational or otherwise dissipated energy.
The simple excess energy measurement of the pendulum does not permit dis-
crimination between these different energy sinks. The separation of the first three
terms becomes possible, however, if the bending force P acting between pendulum
head and specimen is measured (so-called instrumented impact test).
204
II. Deformation, Creep, and l-atIgue ot Unonented Polymers
For the elastic bending of a prismatic beam between supports at a distance Ls

the following expressions for energy and strain are obtained.
The total bending energy We, as a function of beam deflection D, is
(8.12)
the largest density of stored elastic energy in the tensile fiber:
(8.13)
and the largest strain can be expressed as
(8.14)
From this relation the strain rate Em at a pendulum velocity of 2.9 m/s (DIN
51222, 0.5-4 J) is obtained as 65 S-I. The kinetic energy of a sample of a mass of
1 g ejected by the striker at that speed of 2.9 m/s would amount to 4.2' 10- 3 J.
The ratio, an, of the fracture energy An to the sample cross-section B . D is
termed impact strength (Schlagzahigkeit). Such a notation creates the allusion that
an is a fracture surface specific material property. It has frequently been pointed out
that this is not the case [88-89]. Neither We nor Wkin are proportional to the sample
cross-section. A comparison of an values should only be made, therefore, if all values
are obtained from one type of test, preferably even from specimens of identical geom-
etry. Impact strength values an of Charpy un-notched specimens (DIN 53453) at
20°C range from 3.5-12 kJ/m 2 for filled phenol melamine and urea resins,
4-22 kJ/m 2 for various filled epoxy and polyester resins, 12-20 kJ/m 2 for PMMA,
PS, and SAN, and 50-90 kJ/m 2 for ethyl cellulose, CA, styrene-butadiene copoly-
mers, and POM. For many thermoplastics (ABS, CAB, PE, PP, PTFE, PC, PVC, PA)
no fracture is observed under these test conditions [104].
The above impact strength values designate brittle behavior if an < 40 kJ/m 2 .
The materials with an between 50 to 90 kJ/m 2 will generally the brittle if bluntly
notched. Of those polymers not breaking in the Charpy un-notched test some will
respond in a brittle manner if they are sharply notched, whereas others remain tough
even then. Vincent [96] and Bucknall et al. [89] propose, therefore, the following
qualitative impact resistance rating: brittle (an < 40 kJ /m 2 ), brittle if bluntly notched,
brittle if sharply notched, tough but crack propagating, and very tough and crack ar-
resting.
The characteristic impact behavior of brittle polymers can conveniently be out-
lined using polystyrene as an example. Ramsteiner [105] has quite recently performed
instrumented impact test on standard prismatic polystyrene bars. An inspection of
the broken samples revealed that they had failed through rapid propagation of a crack
emanating from the more or less highly crazed tensile zone. The length of the longest
crazes was generally identical to the length of the mirror-like zone of the fracture
surface. The force-deflection curves obtained in this instrumented impact test show
205
a slightly non-linear increase of the load during 1 ms followed by a rapid drop to

zero within less than 50 tJ.S (Fig. 8.24).
The deformation mechanisms taking place during rapid loading can be inter-
preted as follows: The deformation is homogeneous up to a strain level where crazes
are initiated. The craze initiation stress depends on molecular weight, sample treat-
ment, time and temperature (cf. Chapter 9). For the above experiments it would
be reasonable to assume that crazes are initiated once the load attains between 60
and 70% of the maximum load. The maximum of the load corresponds to the point
where rapid crack propagation sets in.
o.92mm
A B
Q.. /
....z
In /'
I /
...
(II
I/'
-
u
0
/
,f .If
.AI .....
deflection 6 deflection 6
Fig. 8.24. Load-deflection curve from instrumented impact test of a brittle polymer (after [1051).
(A) polystyrene as moulded, (B) annealed 30 min at 130 0 C.
The impact toughness of a brittle polymer (area under the load-deformation

curve) is essentially determined by the energy We to reach the stage of rapid crack
propagation (maximum of P). The energy input for separation of the material during
the phase of rapid crack propagation is negligible (Fig. 8.24). One of the most im-
portant aspects of the impact toughness of brittle polymers concerns, therefore, the
limiting conditions under which rapid crack propagation in a material becomes pos-
sible. This problem, the fracture mechanics approach, will be treated in Chapter 9.
The absolute values of the largest bending force Fm or of the largest stresses in the
tensile fiber are not a measure of impact toughness. In fact an annealing treatment
(30 min. at 130°C) of an injection moulded polystyrene bar increased the impact
toughness (due to increased deflection <5) from 18 to 21 kJ/m 2 while decreasing Fm
from 235 to 215 N [105].
Before 'characterizing the role of molecular chains in impact loading the fracture
of normally tough polymers (e.g. HDPE, PVC, PP, PA) will be studied. Cleavage frac-
ture is difficult to obtain by bending and occurs only in notched specimens of low
aspect ratio (Ls/D) or if the rate of loading is high (impact loading). In samples of
HDPE with extremely high molecular weight (Mw > 106 g/mole) cleavage fracture
does not occur at all. The fracture surface shown in Figure 8.25 was obtained by
Gaube and Kausch [106] in impact loading at 20°C from a standard-HDPE bar con-
taining a knife cut. The surface clearly reveals three different morphologies:
a zone of thermally activated crack growth starting from the knife cut (enlarged
in Fig. 8.26)
a zone of unstable crack propagation under a condition of tensile strains
206
a zone where the now rapidly propagating crack is deviated through the stresses
in the compressive fiber of the bending specimen
Within the zone of thermal crack growth two features seem to be of particular
interest, the cellular area II (enlarged in Figs. 8.27 to 8.29) and the void containing
area III (enlarged in Figs. 8.30 and 8.31). Figures 8.27 to 8.31 give an impression
of the large plastic deformation of the surface within rosettes and ridges where the
walls are drawn by up to several hundred per cent. Even the surface of an unstable
propagating crack (area IV, enlarged in Figs. 8.32 and 8.33) is not smooth but plas-
ticly deformed throughout. The morphological details of the fracture surface zones
Fig. 8.25. Fracture surface of notched

standard HDPE broken under impact
loading at 20°C (taken from (106)).
Fig. 8.26. Enlargement of area [ in Fig. 8.25

(taken from 11 06)).
clearly indicate that crack initiation - even under impact loading - does not simply
occur by cutting those molecular chains which happen to traverse the planes of lar-
gest tensile stresses. Due to the limited lateral load transfer between unoriented, un-
crosslinked chains (cf. Chapter 5) plastic deformation always precedes eventual chain
breakage.
207
Fig. 8.27. Enlargement of the cellular area II

in Fig. 8.26 (taken from [106] .
Fig. 8.28. Higher magnification of a section

of area II (Courtesy E. Gaube, Frankfurt-
Hoechst).
Fig. 8.29. Apparently smooth area between the

rosettes visible in Fig . 8.28 (taken from [106]).
208
Fig. 8.30. Voids from area III in Fig 8.26

(Courtesy E. Gaube, Frankfurt-Hoechst).
Fig. 8.31. Higher magnification of the indicated

section in Fig. 8.30. (Courtesy E. Gaube, Frank-
furt-Hoechst).
Fig. 8.32. Enlargement of area IV in Fig. 8.26,

the pJasticJy deformed surface created by an
unstably propagating crack (Courtesy E. Gaube,
Frankfurt-Hoechst).
209
Fig. 8.33 . Higher magnification of a section

of Fig. 8.32. (Courtesy E. Gaube, Frankfurt-
Hoechst) .
On the basis of the foregoing discussion of the phenomenology of impact frac-

ture it can be said that the impact resistance of a material is increased by all molecu-
lar properties or processes which assist the distribution and dissipation of the me-
chanical energy and permit the attainment of a large deflection before inception of
rapid crack propagation. In particular the importance of the relaxation behavior
[91,93-96], of high molecular weight [106], of partial orientation [l08], and of
the unoccupied volume [92] have been emphasized.
A maximum fractional unoccupied volume if) was defined by Litt et al. [92]
through the densities of the amorphous (Pa) and the crystalline state (Pc) as:
(8.15)
The 1 values of common polymers at room temperature vary between zero and
0.12. Litt found a good - although not perfect - correlation between 1 and impact
strength. He indicates that polymers with high impact strength all have 1 > 0.07.
Considerable qualitative evidence has been cited in the literature (e.g. 88- 105)
that the intensities of the low-temperature relaxation peaks deriving from main-chain
motion ({3 and 'Y peaks) are related to impact toughness. A certain correlation has
to be expected since an increased molecular mobility
reduces the level of stored elastic energy at a given sample deformation and load
rate,
prolongs the phase of thermal crack growth, i.e. retards the inception of an un-
stable crack,
and increases the energy expenditure during crack propagation.
Although all three effects tend to increase impact toughness a correlation will
be limited by the fact that the relaxation behavior is determined in the linear anelastic
range of stresses and at small strain rates. These conditions are quite different from
those of an impact test. A quantitative discussion also has to be based on a fracture
mechanical analysis of crack initiation and propagation (cf. Chapter 9).
210
The impact resistance of a polymer is reduced by all factors which lead to a

general or local increase of stored elastic energy at a given deformation but are not
accompanied by an overproportional increase in strength Thus the stress concen-
trating effect of notches, flaws, or inclusions in an otherwise unchanged polymer
considerably lowers its impact resistance. Increasing the degree of cross-linking
beyond the point which insures the loading of all chains only leads to the formation of
short, tightly anchored chain segments. Such segments are the first to be overloaded
and scissioned upon straining. The small impact resistance of fully cured thermosett-
ing resins underlines this point. Reinforcing thermoplastics by short random fibers
increases their rigidity more effectively than their strength which leads to a net decrease
of impact resistance.
A considerable toughening is to be expected, if strength and deformation can be
made to increase simultaneously. Such an effect is realized through the partical orien-
tation of an un-oriented, brittle polymer. For PS drawn to A= 3.4 Retting [108]
reports increases in tensile strength from 47 to 80 MN/m 2 and of strain at break
from 7 to 22%. The IUPAC Working Party "Structure and Properties of Commercial
Polymers" studied systematically the effect of orientation of various polystyrene
samples (homo polymers and rubber-modified polystyrenes) on optical and mechanical
properties [109,110]. It was observed that the impact strength an of unnotched homo-
polymer specimen increased with draw ratio from about 3 at A = 1 to more than
50 kJ/m 2 at A = 6 if the molecular orientation was parallel to the main specimens
dimension. In a perpendicular direction an decreased from 3 to 0.3 kJ/m 2 for the
homopolymer and from about 9 to 2 kJ/m2 for a commercially available rubber-
modified polystyrene [109, 110].
The principal effect of chain backbone properties on impact strength has become
clear. A chain does not so much contribute to impact strength through the level of
stress l/Ib it supports at the moment of chain scission or disentanglement but rather
through the energy dissipated before l/Ib is reached. Chains are loaded in shear which
necessitates the displacement of chains with respect to each other. A maximum of
energy dissipation is obtained, therefore, if the intermolecular shear stresses are large
but insufficient to break a chain (cf. Eq. 5.41) and if the chains do not easily disent-
angle so as to insure chain slip over large volume elements (Fig. 8.28).
B. Failure Under Constant Load
Subjecting a specimen to a constant load evidently can have one of three results:
the specimen breaks upon application of the load
the specimen fails after some time
the specimen carries the load for an indefinite period of time.
Generally the third result is technologically desirable. In order to define the
region of permissible loads a study of the delayed failure, i.e. of the time-dependent
strength, is indispensable.
Owing to its fundamental importance the time-dependent strength of load-bear-
ing polymers has been extenSively investigated with temperature as the principal para-
211
meter. In Figures lA, l.5, and 3.7 stress-time-temperature diagrams of various ther-
moplastics are represented. The phenomenon of a delay between load application
and final failure has found numerous interpretations. Among these interpretations
one group is based on purely statistical considerations. There the time-to-failure tb
is inversely related to the probability that a certain detrimental event takes place
in an otherwise unchanged material Such events are e.g. the formation of a void of
sufficient size or the coincidence of local fracture events. The statistical aspects of
fracture have been discussed in Section II of Chapter 3.
A second group of interpretations is based on reaction rate theory and on the
gradual exhaustion of the polymer load carrying capability due to the breakage
and/or displacement of elements and/or to the formation of defects (cf. Section IV
of Chapter 3). At this point the role of molecular backbone chains in delayed frac-
ture of non-oriented polymers is to be elucidated.
The presence of long and flexible chains causes a polymeric material to show
a monotonic, partly anelastic deformation under constant load, viz. a creep defor-
mation. In statistical theories of fracture generally no special consideration is made
of the extent of creep deformation. From Section IV of Chapter 3 it may be recalled
that the kinetic theory of Zhurkov and Bueche also does not contain creep deforma-
tion as a variable. The theory of Hsiao-Kausch, developed for solids not showing large
scale anelastic deformation introduces a time-dependent strain which derives, how-
ever, as a consequence of the gradual network degradation. Bueche and Halpin ex-
pressly consider the macroscopic creep function to reflect the creep behavior of
molecular filaments which in turn would effect the time-to-rupture. In their theory
the retarding response of a rubbery or thermoplastic matrix causes the delay - by
tb - in the growth of a crack nucleus to a critical size.
The mode of creep failure in thermoplastic materials is not unique. As an example
the creep fracture morphologies of PVC tubes subjected to different stress levels have
been reproduced in Chapter 1. At higher stresses (Oy = 50 MN/m 2) PVC tubes fail in
a brittle manner after little creep deformation, and one speaks of delayed brittle frac-
ture (Fig. 1.1). At medium stresses (42 MN/m2) and after prolonged times the tubes
show large scale plastic deformation, i.e. delayed yielding (Fig. 1.2). At lower stresses
(Oy < 20 MN/m2) failure does either not occur at all within experimental time scales
or by a competing mechanism, the formation of a creep crack (Fig. 1.3).
In brittle fracture the time-to-break tb is determined by the time to initiate a
rapidly propagating crack at a flaw or material inhomogeneity. The role of chains in
this process will be discussed in Chapter 9.
In ductile failure tb is equivalent to the time it takes to develop a local yield zone.
For this to occur it is necessary that the stress level be insufficient to initiate a running
crack at a flaw or material inhomogeneity before the homogeneous anelastic deforma-
tion of the sample volume deactivates possible stress concentrators. A transition from
a ductile to a brittle fracture mode occurs, therefore, if the temperature is lowered
and/or the load rate is increased sufficiently. The structural weakening associated with
the continued creep deformation sets the stage for the eventual local yielding. The·
circumferential creep strain of the afore mentioned PVC pipe at Oy = 42 MN/m 2 is
plotted in Figure 8.34. The curve clearly represents the characteristic parts of a creep
curve: the instantaneous (elastic) deformation EO, a primary phase with a decreasing
212
'" %6 ~
c:: 5 - . '--
'et;
4
--
4 -- --- - _. .
.:.;;-- 1-
:Q 3 ~~ B
C
~
QI
2 V
El
:::J
~ 0 a
u 0 10 20 30 40 50 60 h
time
'" 8 . , . - - - - - - - - - - - - - - , r - .
.c: 0/0
°e 6
t; 5
-0 4
~ 3 Fig. 8.34. Circumferential creep strain E of
Qj 2 a PVC pipe at u y = 42 MN/m 2 and T = 20 DC
E 1 --- (after [1091). (a) linear time scale,
~ O+---,--,.--r-1,----,---.----r----r-rTT~ b (b) E - EO '" (t/h)0.23. A: limit of primary
·u 0 0.01 0.1 1 2 5 10 20 40 60h creep phase, B: inflection point, C: appear-
tim e ance of stress whitening.
strain rate, a secondary phase where the strain rate is constant and a third phase of
accelerated creep. During the latter phase creep rates are large, the material approaches
the yield condition. Such a condition is generally reached first within the most highly
stressed material, i.e. at the smallest sample cross-section.
In order to predict the creep behavior and possibly the ensuing failure a number
of approaches have been proposed. These are based respectively on the theory of
viscoelasticity - including the concept of free volume - or on empirical represen-
tations of e(t) or of the creep modulus E (t) = aojeCt). The framework of the linear
theory of viscoelasticity permits the calculation of viscoelastic moduli from relaxa-
tion time spectra and their interconversion. The reduction of stresses and time periods
according to the time-temperature superposition principle frequently allows estab-
lishment of master-curves and thus the extrapolation to large values of t (cf. Chapter
2). The strain levels presently utilized in load bearing polymers, however, are gener-
ally in the non-linear range of viscoelasticity. This restricts the use of otherwise known
relaxation time spectra or viscoelastic moduli in the derivation of e(t) or E(t).
The limits of linearity are discussed in a review article by Yannas [I 12]. He con-
cludes that for practically all polymers at T-Tg < - 20 K a strain smaller than 1%
forms the limit of linear response. For semicrystalline polymers (e.g. PP, PAN, PETP,
PA 66) a limit of 0.1 to 0.4% seems to be valid also above the glass transition tem-
perature Tg (even at T - Tg > 100 K).
For amorphous polymers (PC, PIB, NR) the limit increases around Tg from 1
to about 50% of strain [112]. There are some special approaches to a calculation
of creep and recovery based on non-linear theories of viscoelasticity [112-114]. At
present it is not possible, however, to predict with sufficient generality the onset of
accelerated creep and thus of delayed yielding from non-linear viscoelastic theories.
213
Promising results have been obtained, however, using the free volume concept to
account for the shift of relaxation times with strain and thermal history and to pre-
dict changes offailure stress and failure mechanism [123,124].
Empirical approaches to represent the creep curves of materials by mathematical
functions have been known for more than 60 years. One of the first is the Andrade
equation
e(t) - €o = {3 (t/tO)1/3 + Kt (8.16)
o
where t 1/ 3 and K are adaptable constants [115]. This equation - originally derived
for metals - describes the uniaxial creep of different materials such as Ph, Cu, Fe,
Cd, PMMA, PA, PVC within certain time periods very well [111, 115, 116]. Various
empirical modifications of this equation have been reported [116]. Thus the circum-
ferential creep is described by a similar potential law
(8.17)
For PVC tubes the exponent n was found by several investigators to be about
0.23 [111]. For the dependency of {3 on 00 a hyperbolic sine function has been
frequently proposed in the literature (e.g. 111, 116). For PVC tubes at 20 °C{3 as-
sumes the form [111]:
{3 = 0.019 sinh (00/8.85 MN m- 2). (8.18)
Equations (8.17) and (8.18) permit the calculation of €(t, 00) at 20 °C. They thus
also permit one to determine at which stress a specified strain value €1 is being reached
after very long times. One finds that a strain value €1 = 3% - at which the creep rate
of moderately stressed PVC tubes (ov = 35 to 48 MN/m2) changes from phase I to
phase II - will be reached at 00 = 22 MN/m 2 only after 50 years. An eventual frac-
ture should take place even later than that.
A still simpler and more accessible approach to predict the creep strain of engineer-
ing components is based on a logarithmicly linear approximation of the creep modulus,
E(t):
log [E(t)/E(t o)] =!.3 log (t/to) log [E(1000 h)/E(l h)]. (8.19)
Values of the 1-h and 1000-h moduli are generally given in the technical literature
(e.g. 104). At 20 °c and at small stresses (5 to 20 MN/m2) the ratios of E (1000 h)
E(1 h) of common polymers are of the order of 0.96 (SB), 0.92 to 0.93 (PC,PETP,
fiber reinforced thermoplastics), 0.88 to 0.90 (PS, PVC, PMMA, POM), and 0.72 to
0.79 (HDPE, PP, ABS). Whereas extrapolation methods are frequently the only means
available they are subject to all the limitations introduced by the gradual development of
structural weakening (demonstrated by the change from a decelerating creep rate to
an accelerating one). Also the action of competing processes (creep crack develop-
ment) is generally not taken into account by this extrapolation method.
214
The development of a creep crack is the dominating failure mechanism of non-

oriented thermoplastics at smaller stresses (see diagrams Fig. 1.4 and 1.5). The frac-
ture surface morphology of a typical creep crack is reproduced in Figure 1.7. The
final failure of that sample had been caused by crack growth over the whole wall
thickness.
The pattern of creep-crack failure shows the three well-known phases: a period
tj of apparently homogeneous deformation up to the point of crack initiation, a
period ta of thermally activated slow crack growth, and a final period tr of unstable
crack growth. An inspection of loaded samples during and after the creep experiihent
reveals a number of morphological features:
creep cracks initiate at flaws or material inhomogeneities, more often in the sur-
face zones (Figs. 1.7, 8.35a) but also in the bulk material (Fig. 8.35b) [111, 117,
118];
Fig. 8.35. Fracture surface of creep cracks in LOPE tubes (Courtesy of P. Stockmayer, [1181).
(a) crack initiation at inside boundary of tube wall, Oy = 2 MN m- 2 , T = 80 °C, tb = 140 h,
(b) crack initiation within tube wall, Oy = 2 MN m -2, T = 80 °C, tb = 7082 h.
the initiated crack grows through thermal activation during a period ta under
apparently constant conditions; this is evidenced by the appearance of the "mirror
zone", the frequently ideally circular or semicircular, smooth interior fracture
surface (Figs. 1.7, 8.35b); depending on the stress level the mirror zone may ex-
tend over almost the complete wall thickness of tubular specimens;
at higher magnifications of the mirror zones it can be seen that the surfaces are
not smooth on a molecular level; in the case of amorphous PVC an irregular, rough,
and scaly surface is observed (Fig. 8.36) and in the case of semi-crystalline PE a
cellular, locally highly drawn surface morphology becomes apparent (Fig. 1.8);
inspection of transparent tubular samples under load and of non-transparent
samples after an interruption of the creep test showed that in PVC and PE creep
crack development started rather late in the experiment, i.e. tj was larger than
0.90 tb [111, 117]; in PP tubes the first creep cracks could be detected at tj being
about 0.5 tb [1171.
Before developing a hypothesis as to the origin of creep crack formation it must
be discussed which information can be obtained from an evaluation of the stress-time-
215
Fig. 8.36. Sections of the mirror zone of a creep crack in rigid PVC (Courtesy E. Gaube, (1071).
U v = 40 MN m- 2 , T = 20°C, tb = 4300 h; (a) Section near the external surface where the crack
originated; (b) Section near the internal surface where the crack terminated .
temperature diagrams (Figs. 1.4,1.5,3.7, and 3.11). The essential feature certainly
is that within a large interval of time the logarithm of the time-to-break tb decreases
linearly with increasing uniaxial or principal stress ao. Anyone of these straight lines
can be expressed by
(8.20)
where to, A, and a are constant with respect to time and stress. As discussed in Sec-
tion IV of Chapter 3 various authors have proposed that these "constants" have a
more general significance. Most approaches work with an Arrhenius equation:
(8.21 )
In this form Uo , 'Y, and to are supposedly independent of temperature. The ex-
perimental data of Savitskii et al. [119] plotted in Figure 3.7 verify this point for a
cellulose nitrate film and the polyamide 6 mono filaments investigated. For each
material only one set of constants is necessary to describe all data points. For cel-
lulose nitrate the authors obtained Uo = 159 kllmol, 'Y = 644· 10- 6 m 3 /mol ,
to = 10- 13 s and for the oriented polyamide 6 (Capron) Uo = 184 kllmol,
'Y = 113 . 10- 6 m 3 /mol, to = 10- 13 s.
In the case of PVC tubes one obtains from an evaluation of Figure 1.4 accord-
ing to Eq. (8.21) values of Uo = 397 kllmol, 'Y = 1740 . 10- 6 m 3 /mol and (a purely
formal value of) to = 1.7 . 10- 52 S. It should be noted that this set of parameters
describes the PVC fracture times notwithstanding the fact that these data belong to
three different fracture modes. In the stress-time curves of non-oriented semicrystal-
line polymers (PE, PP) one observes a drop of strength at larger values of tb, the well-
known bend (Fig. 1.5). The data of the flat portions (relating predominantly to
ductile failure) can be represented by
216
Do = 307 kJ/mol, 'Y = 4390 . 10- 6 m 3 /mol, and to = 3 . 10- 40 S
and the data of the steep portions (creep crack failure) by
Do = 181 kJ/mol, 'Y = 3610 . 10- 6 m 3/mol, and to = 8 . 10- 20 s.
For oriented semicrystalline polymers Zhurkov et al. [18] report:
HDPEDo = 113 kJ/mol, 'Y = 109.10- 6 m 3/mol, to = 10- 11 s

LDPe Do = 117 kJ /mol, 'Y = 247 . 10- 6 m 3/mol, to = 10- 12 S
PP Do = 125 kJ/mol, 'Y = 213 . 10- 6 m 3/mol, to = 10- 12 s.
These formal evaluations need discussion. The good correspondence between

Do and the dissociation energy of the weakest main chain bonds for many of the
(mostly oriented) polymeric materials investigated by him has led Zhurkov [120, 18]
to establish his kinetic theory of fracture. This theory assumes that all important
steps contributing to the fracture of a material are not only related to but even con-
trolled by the breakage of primary bonds, hence by Do and to. As important steps
chain scission, rnicrocrack formation, and crack propagation through material separa-
tion have been considered [3]. The effect on fracture behavior of structure, orienta-
tion, dimensionality of the state of stress, morphology, and homogeneity of chemi-
cally "identical" polymers must be expressed, therefore, through the single parameter 'Y.
The parameter 'Y has the dimension m 3 /mol and is termed activation volume.
With reference to Chapter 3, Section IV, it should designate the thermally activated
volume if the stress 00 indeed acts on this volume directly and without intermediates.
This is approximately the case in flow but not in chain scission. The scission of a
chain necessitates molecular stresses l/Ib which are clearly much larger than the applied
external stresses 00. For PA 6 molecular stress concentration factors l/Ib/OO of
between 10 and 40 are commonly determined (cf. Chapters 7 and 8 above). Taking
12· 10- 6 m 3/mol for the activation volume {3 ofthe chain scission process a param-
eter 'Y = {31/J/oo of 120 to 480· 1O-6 m 3/mol should be predicted. The activation
volume of 12 . 10-6 m 3/mol amounts to (0.27 nm)3 or to a stretching of the ruptur-
ing bond by 0.1 nm.
The effect of the structural parameter chain orientation on 'Y becomes apparent
from Figure 3.11. There the stress-lifetime data of PAN fibers [74] are plotted with
draw ratio Aas a parameter. The highest draw ratio (A = 17.3) corresponds to a'Y of
248· 10- 6 m 3/mol, A= 4 to 590, A= 2.62 to 841 and A= 1 to 'Y = 1200 . 10- 6 m 3/
mol. The 'Y values are the smaller the more homogeneously the macroscopic stress
is distributed over the molecular chains.
The foregoing discussion shows that the 'Y values of PMMA, PVC, and PE can
hardly be explained on the basis of chain scission unless extremely large stress con-
centration values are postulated. In the case of PVC the 'Y corresponds to (1. 7 nm)3
which points to the involvement of larger domains in the activation of the creep
process. The fact that the PVC fracture data of all modes are described by one equa-
tion seems to indicate that all modes depend on the creep deformation. Once a cer-
217
tain structural weakening has resulted from creep the final failure is accomplished
by one of the three modes, e.g. by brittle fracture at large stresses, ductile failure,
or creep crack development. Since the final period accounts only for a small frac-
tion of tb no effect on the slope of Qn tb/tO is noted.
This is different in the case of semicrystalline polymers which form a
"ductile and a brittle branch" of the stress-lifetime curve. There it can be said that
two different mechanisms are active, of which the creep crack initiation has the
apparently smalle~ activation energy (181 kJ/mol) and activation volume (1.8 nm)3.
The fact that chain breakages in PE are rarely observed even at high stresses and low
temperatures in highly oriented samples makes it very unlikely that the mechanism
of creep crack initiation involves chain breakages.
Kagan et al. [121] have studied the effect of supramolecular structure on the
time-dependent strength of HDPE. They found a good correlation between the 1-h
(plastic) creep strength and the yield strength for materials of different crystallinity,
density (0.945 < p < 0.960 g cm- 3) crystallite size, and spherulite diameter. These
parameters hardly affected the activation volume 'Y but slightly the activation
energy of the plastic deformation process (parallel displacement of the ductile branch
of the a-log t/to curve). On the other hand they found a clear tendency to higher
long-tenn brittle strength (resistance against creep crack formation) with an increase
in density and size of crystallites and with a reduction of spherulite diameter. Gaube
et al. [117] also report that an increase in crystallinity (Le. density) of PE, PP, PETP,
POM, PA leads to a higher yield strength - at lower strain values - and to a higher
ductile strength. They point out, however, that the decrease of brittle strength in
the steep region begins sooner, the higher the crystallinity and the lower the molec-
ular weight. The ultrahigh molecular weight polyethylene GUR does not show creep
crack formation at all. Judging from these observations the process of creep crack
formation seems to be related to the gradual disentanglement of chains and the ease
of void opening in the inter-crystalline and/or inter-spherulitic regions. Both mech-
anisms must be fairly independent of creep deformation. The fact that generally creep
cracks are only observable shortly before they cause the final failure indicates that
these cracks grow at considerable rates once they are initiated.
Morphological effects on the initiation of creep cracks were also studied by Stock-
mayer [118]. Using an elaborate peeling technique he transformed whole tube sec-
tions into thin films of up to 0.06 mm thickness (Fig. 8.37). The walls of LDPE tubes,
transformed into films of 0.9 m 2 area, then were scanned under a microscope for
flaws or material inhomogeneities (Fig. 8.38). Inhomogeneities could be detected in
every tube and have been related to an imperfect blending of the LDPE with carbon
black [118]. Stockmayer also found two interesting correlations:
creep cracks were initiated preferentially in those zones which also contained
(the most) inhomogeneities
tube lifetimes were greatly reduced if the creep cracks had been initiated in the
outside or inside zones of the wall; from an ensemble of tubes tested at 80°C
and having an average lifetime tb of 2340 h tubes with inhomogeneities (and crack
initiation) in the boundary zones failed after a it, of 740 h; tubes with crack initia-
tion in the central region of a wall lasted to it, = 7400 h, Le. a factor of 10 longer
[118,149].
218
Fig. 8.37. Part of a film cut from LDPE tube wall showing inhomogeneous blending in extrusion
(Courtesy P. Stockmayer, [118]). Thickness 0.06 mm; width 50 mm.
Comments are necessary concerning the significance of to and 00 in Eq. (8.21).

Values of 10- 12 to 10- 14 s have been frequently determined for to; they have been
assigned to longitudinal vibrational modes of the backbone chain. For PVC and PE
extremely small values are obtained which certainly cannot be interpreted in the
same manner. With flow processes it must be recognized that the activation energy
is not temperature independent but decreases with temperature:
(8.22)
In this case one has
Qn t /t = aUo II T + (Uoo - ,),00) . (8.23)

bOaT RT RT
Fig. 8.38. Section of a film cut from LDPE tube wall showing spots which are free from carbon
black. (Courtesy P. Stockmayer, (1181).
219
A value tg formally determined from experimental data according to Eq. (S.21)

contains, therefore, the two terms to and the temperature-independent part of
aa~o ~~ . The latter term seems to be responsible for tg values of 10- 40 s and less.
The term 00 in Eq. (S.21) has been taken to represent the macroscopic uniaxial
stress 00 in the event of uniaxial stretching - or the largest principal stress component
Oy in the event of multiaxial states of stress. This does not mean that the times to
fracture of tubular and uniaxial specimens are always identical if only Oy is equal 00.
As discussed above the creep functions in these two cases are described by different
potential laws, namely by Eqs. (S.16) and (S.17) respectively.
A systematic study of the effect of the state of stress on the lifetime of PVC
tubes has been carried out by Smotrin et a1. [151]. They found that at short times
to fracture (at stresses of 50 MN m- 2 ) the simple Rankine criterion 0 < 0* repre-
sented their failure data in two-dimensional stress space. With increasing times to
fracture, however, the von Mises criterion gave the better representation. Gotham
[152] studied the uniaxial creep failure of 15 different polymer materials at 20 0 C.
Within a time range of up to 10 7 s he observed brittle failure with injection molding
grade PMMA, PS, SAN, and glass-fiber reinforced PA 66, and ductile failure for PP,
bulk cast PMMA, PC, PSU, PVC, ABS, POM, PA 66, and P4MP.
Concluding this section one may say that there is very little evidence that the
loading and breaking of chains controls the creep of unoriented polymers. The soften-
ing mechanism - common to all linear polymers [122, 12S] - responsible for the
change from decelerated to accelerated creep is generally related to segmental motion
and changes in intermolecular attraction but not to a reduction of chain length or
strength. A search for free radical formation in creep of non-oriented polymers car-
ried out by Jansson et al. [125] is as yet also inconclusive. These authors had included
radical scavengers in PVC which led to a considerable reduction in creep rate (by a
factor of up to about 10). They state, however, that it has not been possible to sepa-
rate the stabilisation effect from the effect of mechanical reinforcement [125].
c. Homogeneous Fatigue
1. Phenomenology and Experimental Parameters
Polymeric engineering materials subjected to strong mechanical and environmental

excitation show - as any other material - a gradual degradation of their performance
including eventual failure. If the property changes are mostly due to chemical reac-
tions one speaks of corrosion, or of radiative degradation. The term fatigue (Zerriit-
tung) is used if a deterioration of material properties is caused by the repeated cyclic
or random application of mechanical stresses. The synergistic interaction of mecha-
nical and environmental attack leads to the phenomenon of environmental stress
corrosion (ESC). The role of molecular chains in ESC, and in corrosive, thermal, and
radiative degradation will be discussed in the final Section of this Chapter. The fa-
tigue of polymeric fibers has already been treated in Sections I F and G of this Chap-
220
ter. The present Section, therefore, will deal mainly with the principal aspects of
fatigue and with isotropic thermoplastics. The fatigue of polymeric engineering
materials has received a growing attention in the last thirty years. Review articles on
this complex subject have appeared from time to time; the reader is referred to the
ones authored by Andrews [126] and more recently by Manson and Hertzberg
[127-128], Oberbach [129-130], and Schultz [131]. In addition, a large number
of research papers have been published of which an incomplete listing is given in
this Section [116,132-147].
It is the general observation that failure in repeated loading occurs at load levels
which are lower than the stresses sustained under the conditions of static loading
(creep) or of monotonously increasing deformation (drawing). The lower the stress
level to which a material is subjected the larger the number N of load cycles which
are sustained. The total time tf, however, which a fatigued sample spends under
load is generally much smaller than the time to failure under static loading condi-
tions. Load reversals or load interruptions have an accelerating effect, therefore, on
the loss of load bearing capability; load reversals or interruptions constitute the ele-
ments of fatigue. It may already be stated at this point that the accelerating effect
of load reversals has to be related to two characteristic material properties:
the anelastic response which leads within each loading cycle to a dissipation of
mechanical energy as heat and
the occurrence of molecular rearrangements which may lead to the disentangle-
ment and slip of chain segments, to local reorientation, and/or to the formation
of voids.
It has to be expected, and must be recognized, therefore, that three principal
mechanisms may contribute to fatigue failure: thermal softening, excessive creep or
flow, and/or the initiation and propagation of fatigue cracks.
The experimental parameters of a fatigue test have been aptly defined by An-
drews [126]: .
a periodically varying stress system having a characteristic stress amplitude, Lla;
a corresponding periodic strain amplitude, Ll€;
a mean stress level, am;
a mean deformation, €m;
a frequency, v; W = 21TV;
a characteristic wave form (sinusoidal, square, etc.) for both the stress and strain;
ambient and internal temperatures, which in general will not be precisely identical;
a given specimen geometry, including notches, if any.
More recently the (computer-programmed) random loading has also found grow-
ing application in testing structural components [132]. Of course, the above para-
meters can be combined in different ways and be changed within a test. They par-
tially depend upon each other and on environmental conditions.
The experimental equipment will not be described here, the reader is referred
to the cited review articles or pertinent research papers (e.g. 132-138). In practice
five different types of stress-strain control are employed which are classified by
Andrews [126] and Manson et al. [127] as:
periodic tensile loading (between fixed stress limits, in tension or compression);
[133-135]
221
periodic loading (between fixed strain limits, in tension or compression); [133-

-135]
reversed bending stress (implicit in flexing a sheet in one dimension); [136]
reversed bending stresses in two dimensions (by rotary deflection of a cylindrical
specimen); [137]
reversed shear stresses, e.g. obtained by torsional deformation. [138]
2. Thermal Fatigue Failure
In cyclic loading of a viscoelastic solid an amount of energy d W is dissipated per

cycle. For sinusoidal loading in the anelastic range, one obtains
u(t) =uosinwt (8.24)

dW =1Tuo€osino. (8.25)
This hysteresis heating leads to a noticeable temperature rise. At low test fre-
quencies and low stress levels the temperature increase of the test specimen generally
approaches a finite value .. A PA 6 sample for instance fatigued at 50 Hz, at a constant
stress amplitude of 8 MN m- 2 , and at an ambient temperature of 21°C, assumed a
stable temperature of 27 °c after some 104 cycles [139-140]. At this temperature
mechanical energy input and thermal energy loss were in equilibrium.
At higher frequencies w, higher ambient temperatures, or higher stress amplitudes,
however, the energy input into a sample becomes larger than the heat which can be
transferred out of the sample through conduction or radiation. Depending on the
experimental setup, failure of the sample then occurs through thermal softening and/
or plastic flow. The stress and frequency range, where this type of thermal failure
has to be expected, can be calculated from the energy input, the geometry, and the
thermal properties of a sample. Thermal equilibrium in a cylindrical sample of radius
R is reached once
(8.26)
approaches zero. The highest temperature, T e , observable at equilibrium in the center

of the cylinder then amounts to
T =T
e a
+[~+ R2] wu~2E' 0 .
2h 4 r..
sin (8.27)
Ta designates the ambient temperature, h the coefficient of convection at the

sample surface; and the constants p, cp ' and r.. have their usual significance as den-
sity, specific heat, and heat conductivity of a material. The term r../pc p is also known
as temperature conductivity number. For conventional polymeric materials this num-
ber assumes values between 1.0 . 10- 7 m 2 /s (PTFE) and about 2.1 . 10- 7 m 2 /s
(HDPE, POM).
222
It should be mentioned at this point that the hysterises heating ~ W in a fatigue

test with constant-strain amplitude decreases with increasing temperature, since ~ W
is proportional to 00 sin 5, i.e. to E". Under this condition thermal equilibrium can
generally be established. The same effect, a decrease of E", can of course be reached
if a sample is plasticized. The plasticization of an engineering component subjected
to fatigue at constant-strain amplitude can be, therefore, a suitable way to prolongue
the fatigue life of the sample [152]. Machyulis et al. [152] point out that the thermo-
stabilization, the fatigue-stabilization, and the plastization effects of various additives
on polyamide are not related to each other. For optimal results on fracture energy
and cycles to failure Kuchinskas et al. recommend diffusion stabilization from a 5%
ethanolic quinhydrone solution at 343 K for 1 h [152].
The conditions of thermal failure are especially discussed by Oberbach[129-130]
for a variety of polymers, by Zilvar [139-140] for PA 6, and by Alf[141] forPMMA
and unsaturated polyester resins.
It is only through their effect on the above constants and on the complex modu-
lus E* that the molecular chain structure influences thermal failure. In the context
of this monograph this type of failure will, therefore, not be discussed any further.
3. Wohler Curves
The common observation that the number of cycles to failure, N F , depends on the
stress or strain amplitude is generally represented in the form of Wohler curves (Woh-
lerkurven, S - N curves). Such a representation accounts for the fact that stress (or
strain) amplitUde and number of load cycles are the most important of the many
(and closely interacting) parameters which affect the fatigue life.
The fatigue data represented by Figure 8.39 were obtained from fully reversed,
uniaxial tension-compression tests at 0.1 Hz with temperature rises Te-Ta below
2 K [142]. The three different regions of the S - N curve of polystyrene are fairly
typical for the fatigue-life curves of other glassy polymers as well. In region I crazes
form during the ftrst tensile quarter cycle. The fatigue behavior is essentially that of
a pre-crazed sample. Failure occurs through the initial slow growth of a crack through
craze material followed by catastrophic crack propagation. The number of cycles to
failure NF is strongly influenced by the stress amplitude, presumably through the
strong dependency of the number of crazes on the stress level.
MNfm2
C1I 50
"U
.....:s I
a.
E
30
0
!II
!II Fig. 8.39. Wohler curve of polystyrene
C1I 10
.....!II
~ subjected to fully reversed uniaxial tension-
0 compression at 0.1 Hz at ambient temper-
0 ature (after [1421). NF number of cycles
to failure.
223
Region-II fatigue fracture is characterized by the fact that a period of craze nu-
cleation precedes craze growth and the formation, slow growth, and catastrophic
propagation of.a crack. This type of fatigue fracture is observed at stress levels
just below the immediate craze initiation stress ai. The dependency of NF on a is
considerably reduced. This accounts for the delay in craze initiation as well as for the
reduced rate of slow crack growth at lower stress levels. The slope (~ 14 MN m- 2
per decade of NF) seems to be characteristic for a variety of polymers [142, 153].
The long-life region III essentially forms the endurance limit of the material.
The dependency of NF on a is extremely small. Crack initiation has a very long in-
cubation period. Analysis of the fracture surfaces of fatigue failures in this region
reveals furthermore that the slow crack growth mode has two phases. The first shows
little evidence of crazing, the second resembles the region-II failure mode.
The above considerations concern polystyrene fatigued by completely reversed
stress (am =0, I:l.a = 2a). In principle similar observations are also made with other
polymers and at other conditions of stress, e.g. under constant maximum stress at
various stress amplitudes [127, 153]. At this point is should be emphasized that the
"fatigue process" under conditions II and III comprises at least two phases: an initial
phase of crack or craze initiation with an apparently homogeneous response of the
material and a second phase of heterogeneous fatigue with localized and growing
material discontinuities (fatigue cracks). The second phase in turn comprises several
modes of crack propagation (thermal and athermal growth, catastrophic failure).
This will be discussed in more detail in Chapter 9.
The discrimination between homogeneous and heterogeneous is certainly useful
although not necessarily always significant. Most of the kinetic theories of failure
treated in Chapter 3 encompass the two phases of crack initiation and slow crack
growth. Thus they implicitly assume that the entity of molecular rearrangements
between the application of load and the end of the slow crack growth period can
be described by one constitutive equation.
4. Molecular Interpretations of Polymer Fatigue
Different experimental observations permit the conclusion that the prolonged peri-
ods of craze or crack initiation at lower stresses are not simply due to the reduced
probability for crack nucleation in an otherwise unchanged material. The nature of
the changes occurring on a molecular level during sample fatigue have been investi-
gated by several authors [e.g. 138,143-147,153]. Thus molecular mobility, inter-
action, and damping have been studied by the torsion pendulum [138,134-144]
and by infra-red technique [138], molecular packing and defect structure by density
measurements and X-ray scattering technique [144-146] and by the use of samples
of different molecular weight [153], and morphological changes by the recombina-
tion kinetics of trapped free radicals [146]. The results obtained from these different
experimental methods characterize the molecular rearrangements, but they do not
yet permit the establishment of quantitative fatigue limits.
Through simultaneous application of several of these methods Sikka [144] has
been able to characterize some molecular rearrangements in homogeneous fatigue.
He fatigued in cyclic extension thin films (0.075 mm thick) of polystyrene (Tricite)
224
II . Deformation, Creep, and Fatigue of Unoriented Polymers
and polycarbonate. Subsequently he investigated these films by Fourier transform

infrared (FTIR) and mechanical spectroscopy and by X-ray diffraction technique.
The fatigued PS samples were scanned under an electron microscope to detect any
crazes that might have developed due to fatigue . No crazes were observed in the
samples fatigued up to 2500 cycles.
The FTIR results obtained on a PS film after 2500 cycles are shown in Figures
8.40 and 8.41. In the superimposed FTIR spectra slight distortions of IR bands
are noticeable in the frequency range of 900 to 1700 cm-' (Fig. 8.40). To detect
minute changes caused by the fatigue process Sikka also recorded computer-sub-
1.25
1.00
'"
v
c 0.75
0
J:J
~
0
III
J:J
0 0.50
0.25
0
1700 1500 1300 1100 900 em-1
Fig. 8.40. Superimposed FTlR spectra In of unfatigued ( - ) and If of fatigued (---) poly-
styrene film (from [144 I).
0.15
0.10
'v"
C
o
-eo
III
-g 0.05
O~~~-W~~~~--~~~__~__-L~-L~
1600 1400 1200 1000 800 em'
Fig. 8.41. FTlR subtraction spectrum of polystyrene, If-0.9 In (from (1441).
225
tracted FTIR spectra. One such spectrum (fatigued-90% unfatigued) is plotted in

Figure 8.41, the changes observed in this spectrum are listed in Table 8-IV.
From X-ray diffraction data Sikka obtained a mean Bragg distance d Bragg of
about 0.48 nm for the unfatigued PS films which showed a decrease of ~ 0.01 nm
in the fatigued sample (2500 cycles). This negative shift in d Bragg was believed to be
conclusive. It was related to the decrease in interphenyl and intraphenyl distances.
The fatigue to 2500 cycles also resulted in a change of dynamic mechanical losses
[144). Figure 8.42 shows a plot of tan 8 versus temperature obtained at a frequency
of3.5 Hz.
Table 8.4. Fatigue-affected IR Bands of Polystyrene (from Sikka [144]).
IR Band Frequency Observation

(em-I)
540 Multiple splittings

906 Part of the band shifts to lower frequency at 888 em-I. Shift amount
is -18 em-I.
1154 Part of the band shifts to lower frequency at 1145 em-I. Shift amount
is -9 em-I.
1310 Part of the band is contributed by fatigue process.
1376 Part of the band appears at a higher frequency of 1400 em -1. Shift
amounts to 24 em -1. Rest of the band is resolved into bands at
1352, 1358, 1365, 1371, 1376 em-I.
1448 Shows strong splitting.
1478 Two additional bands appear at 1489 and 1509 em-I.
2850 Weak splitting observed.
3062 Medium intensity splitting observed.
3083 Medium intensity splitting observed.
10
,. 8
...
!:
.,
~
6
2
-- -- Fig. 8.42. Tensile mechanical loss tangent
o L.....I_ _" " - _ - ' -_ _' - - _........_........... of polystyrene at 3.5 Hz (from 11441).
-100 -60 -20 20 60 100 1. unfatigued sample; 2. sample fatigued
TEMPERATURE (DCI to 2500 cycles.
226
The FTIR results seem to indicate that the fatigue process has caused changes
in the absorption frequencies of various vibrational modes of the phenyl side groups.
Also the IR bands associated with bending and stretching modes of CH 2 groups and
bending modes of CH groups show frequency shifts and band distortions. These
results suggest that fatigue has modified the internal and external environment (in-
trachain and interchain interactions) of PS chains.
On the basis of these assignments and of his X-ray studies Sikka proposes that
during fatigue weaker Van der Waal's and possibly some backbone bonds are broken.
This has a twofold effect. The loosening of the structure permits on one hand the
PS chains to locally rearrange themselves in a more perfect manner and to reduce
the average intraphenyl and interchain distances. The local rearrangements are, on
the other hand, accompanied by an increase in free volume in interstitial areas be-
tween the domains of improved order.
In his experiments with polycarbonate films Sikka [144] used cyclic tensile
stresses between 50 and 91 % of the yield stress. In the upper stress range the stress-
strain curve was clearly non-linear. He employed the same experimental techniques
as mentioned before to study molecular fatigue effects. The (occasionally minute)
shifts in the IR spectrum are made apparent in the computer-subtracted FTIR spec-
trum (Fig. 8.43). Sikka's vibrational assignments of the affected bands are compiled
in Table 8.5. They indicate a stretching of some backbone chains. X-ray diffraction
again revealed change of the mean interchain distance. It amounted to a decrease
of 1.25% after 5800 cycles under a tensile stress reaching 83% of the yield stress Oy .
No effects of fatigue on orientation could be detected in such samples by birefrin-
gence measurements. The densities decreased by a fraction of up to 0.083% to
1.1966 g/cm 3 .
......
<"-
;::
......z
........
CIC
0
'"C> ~
~
'"<"-~ '"
N
§
'"'"'" '"
'"
<"-
3U. n oa 1600 140 0 12 0 0 800 600
WAVENUM B ERS CM - 1
Fig. 8.43. FTIR SUbtraction spectrum of polycarbonate, If - 0.9 In (from [1441).
227
Table 8.5. Changes due to Fatigue in IR Bands of Polycarbonate (from Sikka [144])
Type of Changes Assignment

Observed
560 Multiple splitting Out of plane bcnding vibration of the

p-disubstituted ring
768 Splitting Non-coordinated bands
886 Appearance of a shoulder Non-coordinated bands
1012 Multiple splitting A 1 species modes of the benzene ring
1078 Splitting Asymmetric C-C stretch of central
carbon atom
1150- 1268 Absorbs on the negative C-O valence vibration of
absorbance side \\C-O-C-O- C1/ group
/ II \
o
1362 Weak splitting Aliphatic -CH valence vibration
1408 Very weak splitting Non-coordinated bands
1502 Multiple splitting In plane ring vibrations of the phenyl
rings
1775 Multiple splitting C=O stretching vibration
2968 Splitting conformation around central C atom
of the repeat unit has an effect on this
band.
Some dynamical mechanical loss data are represented in Figure (8.44). The
Q and ~ relaxation peaks in fatigued samples appear at lower temperatures in com-
parison to the corresponding peaks in unfatigued samples. In all cases the samples
fatigued to large numbers of cycles (> 2000 cycles at stress levels umax/u y > 0.56)
fractured during the dynamical mechanical tests.
12 CURVE 1 • COH TROl

l . 1J3I CYCLES
J = lDDD eyc l ES
10 4 = lDDD eye lE S
.....
1
~ 8
.
""c:
4
o~~~~~~~~~~~~~~~~ __~
-160 -120 -80 -40 0 40 80 120 160
TEMPE RATUR E lOCI
Fig. 8.44. Tensile mechanical loss tangent of poly carbonate at 3.5 Hz (from [1441.
228
Also in this case one has to conclude that under fatigue stresses chains are
stretched and locally packed more closely whereas the overall free volume increases.
The observations discussed in detail above seem to be representative for the
findings of other authors as well. Bouda et al. [138] report a decrease of density with
cycle number for PA 6 and PMMA which they ascribe to the formation of micro-
voids. The decrease of heat capacity in PMMA in the temperature range 350 to
400 K, the decrease of tan 8 at 93 K (1 Hz) and an increase in shear modulus G' in
the range 100 to 250 K (where no crazes had formed) seemed to indicate a non-
uniform local volume contraction during the fatigue process.
Schrager [134] investigated epoxy resins of different degrees of cure. A fully
cured sample showed a monotonous decrease of G' and an increase in tan 8. Notable
is a complete and persistant recovery of G' and tan 8 within 20 h. An incompletely
cured sample responded to fatigue loading by an initial rise of G' (indicating comple-
tion of chemical cure simultaneously to chain scission in other parts of the matrix).
After some 20· 10 3 cycles chain scission and a decrease of G' dominated.
Wendorff [145] studied fatigue-induced density fluctuations in polyoxymethylene
(Hostaform T 1020) by X-ray scattering. He reports several interesting structural de-
tails of these defects:
a constant concentration of about 10 12 cm- 3
a size growing with stress amplitude and cycle number
a shape resembling that of a rotation ellipsoid (with axes 2a = 2 to 6 nm, 2b =
9 to 50 nm)
the defects have a compliance some 3 to 5 times larger than that of the bulk
material
the defects show viscoelastic behavior and a size distribution
annealing close to the melting temperature leads to partial healing; the defects
reappear if cycling is resumed
form and size of the defects depend on crystallization conditions.
These observations indicate that the defects are microvoids appearing preferen-
tially at interlamellar boundaries oriented perpendicular to the loading direction.
Similar defects are observed in homogeneous deformation during static loading at
those strains where the stress-strain curve begins to deviate from linearity [145].
Whereas Wendorff states that fatigue does not affect the crystalline regions of
POM Nagamura et al. [146] report a change of the crystalline mosaic block struc-
ture of HDPE. They arrived at this conclusion by analysis of the trapping and decay
behavior of 'Y-ray induced free radicals.
Measurable macroscopic parameters which have been related to the progress of
material fatigue are creep strain and strain rate [72, 116, 122, 123, 147]. Mindel et al.
[122] studied the creep rate as a function of strain in pure compression of polycar-
bonate. They found that fatigue loading becomes effective through an increase of
strain rate after each interruption of loading. Since the strain level after which ac-
celerated creep began remained constant (at 8.8%), fatigue lives were reduced. Ten-
sile creep has frequently been held responsible for the fatigue failure of polymers.
In 1942 Busse et al. [72] suggested this mechanism for polyamide, cotton and rayon.
Briiller et al. [147] stated cyclic creep strains were predictable from the Boltzmann
su perposition principle.
229
Sauer et al. [153] studied particulary the effect of molecular weight on fatigue
of PS and PE under different alternating stresses. They observed that an increase in
molecular weight dramatically increases the fatigue life of PS (an increase of the
number of cycles by a factor of 10 for an increase of M by a factor of 5). This effect
is partly assigned to the retardation in craze initiation because of the fewer number
of chain ends and the larger number of entanglements in high molecular weight
samples. The major contribution derives, however, from the increased resistance of
the crazed material to rupture (cf. Chapter 9.11).
Comparing the fatigue behavior ofPE and PS Sauer et al. [153] conclude:
in both crystalline polymers, like polyethylene, and amorphous polymers, like
polystyrene, increases of molecular weight lead to increased fatigue lifetimes, for
a given imposed alternating stress, and to increased endurance limits,
the ratio of the estimated fatigue endurance limit strength to the nominal static
fracture strength for specimens prepared from narrow molecular weight poly-
styrene standards rises appreciably with increasing molecular weight, from a value
of 0.19 for a molecular weight of 160,000 to a value of 0.30 for a molecular weight
of2·106 .
if fatigue failure due to excessive thermal heating is avoided, the resistance of
polyethylene of medium or high molecular weight to alternating loading is con-
siderably greater than that of polystyrene, based on static stress values to produce
either yield or fracture. Thus, at comparable molecular weights of about 2 . 106 ,
polyethylene gives an estimated ratio of endurance limit strength to yield stress of
about 0.90 as compared to the 0.30 ratio for polystyrene,
the greater resistance of crystalline PE to fatigue type failure as compared to
amorphous PS is attributed primarily to two characteristics. The first concerns
the phase of homogeneous fatigue. Because PE has a much lower amorphous-
phase Tg and consequent greater molecular and chain mobility of its amorphous
chains, it is much less sensitive to surface or volume flaws; and secondly, because
of its heterogeneous nature, with crystallites imbedded in an amorphous matrix,
it is also much more resistant to fatigue crack propagation (cf. Chapter 9).
D. Yielding, Necking, Drawing
Shear yielding, i.e. the onset of large-scale intersegmental displacements within a non-
oriented thermoplastic polymer, has a distinct effect on the stress-strain curve. In ten-
sile tests a drop in engineering stress is usually observed and the yield point defined
as the point of maximum load (Fig. 2.10, curves b and c). In other tests, e.g. in a com-
pression test, there mayor may not be a load drop but a sudden decrease of do/de
may be noticed. The important phenomenon of yielding has been intensively inves-
tigated. Review articles have appeared almost annually in recent years in the general
literature [e.g. 114,154-164].
The classical continuum mechanical criteria for shear yielding have been dis-
cussed in Chapter 3, Section III A in terms of the three-dimensional state of stress.
At this point molecular interpretations are reviewed in an attempt to assess the role
of molecular backbone chains.
230
Experiments at different temperatures, strain rates, and hydrostatic pressures

have established that shear yielding is a thermally activated process [154-168,
170-173]. According to Eyring's theory of flow (Chapter 3) strain rate can be ex-
pressed as
€ = A exp - (Vo - ru)/RT. (8.28)
This gives the following dependence of yield stress u y on strain rate and tem-
perature:
uy Vo R €
- = - +-Qn- (8.29)
T rT r A·
This equation was found to be valid for a number of polymers (pVC, PC, PMMA, PS,
CA) in more or less extended regions of temperature and strain rate [154,156,158].
The (temperature-dependent) activation volumes r had at room-temperature values
between 1.4 (PMMA) and 17 nm 3 (CA). This means that according to this concept
polymer deformation at the yield point is due to the thermally activated displace-
ment of molecular domains over volumes which are between 10 (PMMA) and 120
times (PVC) as large as a monomer unit. It has been indicated by several authors
[155-158,160] that the above criterion (Eq. 8.29) corresponds to the Coulomb
yield criterion TO + IlP = constant. The coefficient of friction 11 is inversely propor-
tional to r. From an analysis of their experimental data on polycarbonate according
to Eq. (8.29) Bauwens-Crowet et al. [158] conclude that two flow processes exist.
They relate these to an a-process Gumps of segments of the backbone chains) and
to the ~ mechanical relaxation mechanism.
Argon and Bessonov [161,162] have recently presented a molecular model of
shear yielding which involves the rotation of molecular segments against intra- and
intermolecular forces and the systematic, pairwise reduction of kinks in a strained
polymer. They calculate the free enthalpy of activation of a double kink in a small
bundle of collectively acting molecules, LlG*, as
LlG* = 31TGw 2a 3 '[I _ 8.5 (1 _1)5/6 (~) 5/6] , (8.30)

16(1-1) G
where G is the shear modulus, I) Poisson's ratio, w the angle of bundle rotation dur-
ing the activation, and a the mean bundle radius. Having examined a large number
of glassy polymers Argon and, Bessonov fmd that the bundles comprise between
about one (PMMA, PS), two (PPO), three to four (PC, PETP) or seven (PI) molecular
chains. The distances between double kinks along the molecule range from 0.8 to
1.6 nm [162].
All of these proposed interpretations of yielding state that yielding is brought
about by a displacement of neighboring chain segments under change of conforma-
tion. In the process of yielding of unoriented thermoplastics no large axial chain
stresses are built up and no chain scission has ever been observed in strains below
the yield strain €y. Yielding is equivalent to the onset oflarge scale orientational
deformation. It is generally accompanied by a decrease of the material resistance
231
against deformation, by a reduction of sample cross-section in a plane perpendicular

to the direction of tensile plastic deformation and by a temperature increase due to
the dissipated mechanical work. Material and thermal weakening at constant or in-
creasing true stress give rise to a plastic instability. In uniaxial tensile straining this
instability manifests itself as
shear banding [155-157],
neck formation [165-175], or
homogeneous drawing [176-182].
The phenomenon of crazing also can be regarded as a (although localized)
deformational instability. It will be discussed in the final chapter.
The large scale "plastic" deformation associated with yielding quite naturally
gives rise to shear and extensional deformation of molecular coils.The response of
molecular chains in solution to shear or extensional flow has been described in Chap-
ter 5I1E. It has been stated that chain breakages are obtained if strain rates are suffi-
ciently large (generally above 600 S-l). The strain rates in a yielding solid are con-
siderably smaller. The local strain rates across the neck of a sample are typically 1 s-l
or less (Fig. 7.12). On the other hand molecular friction coefficients in a solid are so
much larger than in a liquid that large forces will be transmitted once a chain mole-
cule slips through the surrounding matrix. As discussed in 5I1 E axial forces sufficient
for chain breakage will be built up if the slipping chains have a highly extended con-
formation.1t must be expected, therefore, that all those chain segments rupture as a
consequence of necking or drawing which are properly oriented and which initially
do have a highly extended conformation or assume such a conformation at the early
stages of the yield process. With regard to chain breakage in necking or drawing the
total orientational strain A. and the drawing temperature T d are the most important
variables.
There are very few reports on ESR investigations of chain rupture during necking
and drawing [21, 169, 174-177]. These reports will be discussed in the following and
be supplemented by observations with other methods: the IR spectroscopy of chain-
end groups [178] and the determination of molecular weight distributions [179].
Zaks et al. [169] investigated the neck formation in polycarbonate. As inferred
from the reduction of sample cross-section in the neck, an orientational strain A. of
about 2 had been imparted onto the material while "passing through the neck". At
room temperature and at various sample stretching rates corresponding to strain
rates in the neck of between 0.02 to 2 S-l , the authors [169] observed a fairly stable
but not very well resolved ESR spectrum. The intensity of this spectrum increased
with the rate of passage of undrawn PC through the neck from 3 . 10 15 to 1.8 . 10 16
spins/g (Fig. 7.12). The authors also studied the behavior of stable nitroxide radicals
and of radicals formed by photolysis during necking of LDPE and PC. The thus ob-
served increased decay of the initially present radicals may be assigned to the reac-
tion with newly formed radicals but, as well, to the strain-induced increased rate of
recombination or decay of radicals already present. The existence of the latter phe-
nomenon in highly oriented HDPE, PP, PA 12, and PETP was pointed out by Becht
et al. [47].
The ESR technique had also been employed by Lishnevskii [174, 175] who found
that during cold drawing of PP, PVC, PA 6, and aromatic polyamide fibers free
232
radicals were formed in a concentration of between 8· 10 14 to 3· 10 16 g-1.

Matthies et al. [176] noted the intense coloring of different polyoxamide and hexa-
methylene isophthalamide fibers during cold drawing. They confirmed by ESR measure-
ment that the formation of free radicals had given rise to the appearance of the color
centers. In the case of the polyoxamide fibers they associated these color centers
with the -NH-CO-CO-NH-group. The largest concentration of spins observed
was 1 . 10 16 g-1.
The above ESR experiments [174-176] were carried out on cold drawn
samples immediately after completion of the drawing procedure. Thus the measured
radical concentrations [R] are integral values. Nothing can be said about the form of
[R(A)], but, in all probability, most of the chain breakages have occurred during the
final stages of the drawing procedure, i.e. within moderately to highly oriented and
even microfibrillar structures. This assumption is supported by the ESR experiments
of Levin et al. [21] and Chiang and Sibilia [177] with fibers partially oriented at ele-
vated temperatures.
The hot drawing of PA 6 at temperatures of e.g. 150 to 210°C permits one to apply
orientational strains of up to A= 6 without breaking an observable number of chains
[21]. As discussed in detail in Chapters 7 and 8-1, free radicals are formed if such fibers
are stretched at room temperature. For a PA-6 fiber hot stretched to A= 3.7 radical
formation began at an "additional cold" strain of 20%, i.e. at a total elongation of
A= 4.44 [21]. PETP fibers hot stretched to A= 3 showed some radical formation
at "cold" strains of between 5 and 20% but deformed by chain slippage thereafter -
up to a strain of 50%, i.e. to a total elongation of A = 4.5 [177]. In HDPE films no
breakage of chains could be observed during cold drawing unless the draw ratio A
exceeded 5 [178].
An infra-red study of the latter material also revealed [178] that the number N
of coiled rotary isomers (chains containinggg andgtg conformations) depends on
the draw ratio A. While N (A) is rather constant in < 5, N decreases linearly at higher
draw ratios to reach about 0.2 N (1) at A= 12.
That chain scission occurs during drawing of PA-6 multifilament yarn becomes
apparent from an investigation of the molecular weight. Using a fractional precipi-
tation technique Sengupta et al. [179] determined the molecular weight distribu-
tions of undrawn yarn and of yarn drawn at 35°C to A = 2.92 and A = 4.20 respec-
tively. They observed that for the yarn drawn to A = 2.92 the molecular weight
distribution is unimodal as in the undrawn fiber, but the peak is smaller and
slightly shifted toward smaller values of M. At A = 4.2 this peak is further di-
minished and shifted and a second peak at low molecular weights (at less than half
the viscosity number) appears [179].
Indirect evidence of the likely occurrence of chain scission was proposed by
Davis et al. [170-173] who studied the deformation behavior of high molecular
weight low density PE (and of PTFE) at hydrostatic pressure. From the pressure
sensitivity of the rate of plastic deformation they derived a zero-pressure activation
volume of 0.266 nm 3 which they suggested to correspond to the activation volume
'Y for the breakage of a PE chain.
From the presented evidence it must be concluded that during the initial stages
of necking and drawing (A < 3) of semicrystalline polymers chain breakages do not
233
occur in large numbers. This conclusion is completely in accord with the Peterlin
model of the "plastic deformation" of semicrystalline polymers. As described in
2.11 B the early stages of plastic deformation comprise the destruction of an original
spherulitic texture and its transition to the new microfibrillar structure. During the
transition interspherulitic tie molecules are most highly loaded and some are ex-
pected to break. The ties between the crystal blocks, however, are generally not
severed at this stage and are instrumental in establishing the microfibrillar structure.
If a semicrystalline polymer is carried from moderate to high extension ratios
(3 < A. < 10) then chain breakages can occur [21,169,174-178]. All of the men-
tioned breakages must be considered to be of the static-loading type (cf. 5 II Band
D). They become possible in large numbers because the presence of laterally rigid
crystalline regions permits:
the static transmission of large axial stresses onto chains,
the balancing of ensuing irregularities of stress distribution, and thus
the widespread attainment of critical chain stresses before macroscopic break
down.
In recent years ultra-high modulus fibers with draw ratios up to 35 have received
increased attention [180-182]. Their highly extended chain structure and their ex-
treme mechanical properties (elastic modulus E up to 67 GN m- 2 , ab up to 1.7 GN
m- 2 ) have been discussed in 2IC, and the techniques for their processing (drawing,
hydrostatic extrusion, solution spinning) are outlined by Bigg [180]. As yet there
is no information on the extent of chain breakage during drawing or stressing of these
fibers. Capaccio et al. [181] point out that in very highly drawn PE (A. = 30) much
finer fibrils are present and these are characterized by an amazingly high concentra-
tion of longitudinal discontinuities in the form of microfibril ruptures. Molecular
defects (chain ends, folds) are certainly present in extended chain crystals - but not
in quantities indicating major chain breakages during processing [180-182].
With the exception of PC non-oriented amorphous polymers have not given rise
to measurable quantities of broken chains during yielding in tension. This behavior
is the result of their different morphology. In the absence of crystallites, large axial
rorces leading to chain scission can only be caused by frictional loading of slipping
chain segments. The volume concentration of chain breakages (because of the larger
size of the slipping segments) is predictably much smaller than in semicrystalline
polymers. In addition (and because of the absence of the equalizing effect of a
microfibrillar substructure) macroscopic tensile failure begins before stresses and
strains have reached a level sufficient for widespread chain scission.
During necking or homogeneous drawing of many (transparent) polymers an
intensive whitening is observed. The origin of this phenomenon are voids which are
either formed in a correlated manner within crazes (cf. 9 II) or within shear bands
or which appear in an uncorrelated manner within the deforming volume elements.
Uncorrelated voids form in semicrystalline polymers (PE, PP) as well as in amorphous
ones (PVC) and in elastomers at cryogenic temperatures (BR, PI). Figure 8.45 shows
an electron micrograph of a cross-section through the voided area which appeared
during necking of a dry PA 6 monofilament [83]. The fact that the voiding starts in
the center of the necking region and in dry monofilaments only, indicates that it is
related to a triaxial state of stress. The coalescence of such voids may give rise to the
234
II. Deformation, Creep, and Fatique of Unoriented Polymers
Fig. 8.45 . Electron Micrograph of ultrathin cross-section through voided area of drawn 6 poly-
amide monofilament (Courtesy H. Hendus, (831). Magnification 4800 x.
fibrillar fracture shown in Figure 8.22 [83]. From the triaxiality of the initiating
stresses and from the absence of measurable intensities of free radicals in slightly
drawn isotropic polymers (A < 3), from the absence of conspicuous molecular weight
changes [179,220], and from the reversibility of the formation of holes it must be
concluded that the hole formation depends more on conformational rearrangements
of chains and on surface free energy than on the breakage of chains .
The material behavior described above referred mainly to uniaxial tensile defor-
mation. It will be different in techniques of working or forming of polymeric mate-
rials, such as rolling, sawing or grinding. These processes involve compressive, tensile,
and shear deformation in a generally complex manner. Breakage of chains has been
observed in various ways. In 6IV Band 7IC3 a detailed account of the nature and
concentration of free radicals formed in comminution has been given. Other me-
chano-chemical aspects will be discussed in 9 III D.
E. Elastomers
It is evident that the role of highly coiled, loosely interacting but permanently cross-
linked chain backbones in the rupture of elastomeric materials must be quite different
from the role of chains in thermoplastics. The basic principles and notations of rub-
ber elasticity have been presented in Chapter 2 (Section IIA), different fracture
models in Chapter 3, and a discussion of the entropy-elastic deformation of a single
chain in Chapter 5 (Section IA). If one speaks of a "chain" in relation to a filled
and cross-linked polymer network one refers, of course, to the part of a molecule
between adjacent points of attachment (filler particle or cross-link). Thus chain rup-
ture refers to the breaking of a chain at or between these points. It has been proposed
by Mullins [183] some thirty years ago (and has since been known as Mullins effect)
that the breakage of chains in a first loading cycle is responsible for the softening ob-
servable in subsequent load cycles. In that respect the phenomenology of the Mullins
effect is comparable to that of the breakage of chains in the stretching of fibers.
235
The values of the entropy-elastic forces acting on highly extended 5-nm chain
segments (cf. Table 5.2) indicate that they are by more than one order of magnitude
too small to cause main chain bond breakage. For a chain of 125.5 nm length break-
ing stresses are reached if the chain is extended to an end-to-end distance of 124.7 nm
which corresponds to a stretch ratio A. of 18. Rupture of most technically important
elastomers takes place, however, at stretch ratios A. smaller than 8 [183-195]. As
shown in Figure 3.6 and indicated in the literature the reduced ultimate stresses ub
plotted versus elongation A.b form a failure envelope. The elongation to break A.b is
shifted along the failure envelope (towards larger elongations) if the temperature
and/ or the degree of cross-linking decrease or if the strain rate increases [183 -19 5].
The macroscopic breaking stresses ub are of the order of 1 to 30 MN m -2 (depend-
ing on degree of filling or cross-linking), they decrease with increasing time-to-failure
and degree of swelling.
From these observations it has been quite unanimously concluded that an aver-
age network chain influences the deformation and breakdown of a stretched rubber
in several ways:
the originally random chain conformation changes to a more extended one at a
rate determined by strain, temperature, and internal viscosity,
in highly cross-linked or filled rubbers numerous (of the shortest) network chains
have been completely extended and broken at the fracture elongation (cf. 7 II),
additional physical cross-links are introduced through crystallization at high
elongations,
voids or cavities can be formed due to non-homogeneous distribution of strains
in filled or unfilled elastomers,
the limits of extensibility are reached once the majority of the network chains
has reached a highly extended conformation. In the case of a homogeneously
cross-linked network with n random links per network chain the largest breaking
elongation A.b max will approach Vn. At that strain all chains oriented in stretch-
ing direction are fully extended (cf. 5 I A).
Studying the influence of the aforementioned parameters on the load bearing
capability of elastomeric networks one may say that forces are transmitted
through the entropy-elastic deformation of network chains between cross-
linking points,
in the case of highly extended chains through their energy-elastic deformation
(cf. 5 I),
through shearing forces between segments.
The first mode of load transfer prevails in the rubber-elastic time-temperature
region, and the third at very low temperatures [190].
For the breakdown of a stressed rubber two basic mechanisms are conceivable:
the initiation and growth of a cavity in a moderately strained matrix or the accelerat-
ing, cooperative rupture of interconnected, highly loaded network chains. It has
generally been concluded that the second mechanism is of dominating importance
under experimental conditions, permitting the largest breaking elongation A.b max to
be attained [187, 192, 194, 195]. For a network with chains extended in loading
direction a dependence of A.b max on the inverse square root of the cross-link density
lle is predicted. In experiment A.b max was found to be proportional to (lle)-O.5 to
236
III. Environmental Degradation
(V e)-U.78 for a variety of networks (Butyl, BR, EPR, SBR, SI, fluorinated rubber).
Smith [187] concludes that the maximum elongation Abrnax is the only feature of
large deformation behavior which clearly depends only on network topology.
At ultimate strains which are smaller than Abmax some chains are already highly
extended but their number is insufficient for cooperative accelerating breakdown.
It is the general opinion that under such experimental conditions the final break-
down occurs through formation and growth of cavities [184,185, 190]. It will not
be further investigated at this point to what extent chain scission may accompany
cacity formation and growth. It may be stated, however, that the chain scission
mechanism does not determine the rate of cavity formation.
As a final point it should be emphasized that elastomeric materials are showing
a higher strength if they exhibit a multiphase microstructure [183, 184, 186, 188,
195]. This may be achieved by incorporation of suitable filler materials (carbon
black, silica); by crystallization under strain, or by blending or copolymerization
with an incompatible polymer. The possible role of chain orientation, loading, and
scission in these cases has been discussed in 7 II.
Within the framework of this book the effect of thermomechanical chain scission on
the mechanical properties of polymers are to be investigated. Up to this point it has
been tried, therefore, to separate and exclude environmental effects whenever pos-
sible. In many cases it was tacitly assumed that the variables investigated (e.g. strain,
stress, temperature, sample morphology, free radical concentration) had been the
dominating variables in comparison to e.g. humidity, oxygen content, chemical at-
tack, or effects of irradiation. It goes without saying that these environmental fac-
tors are of extreme importance to the service lives of polymer engineering compo-
nents. A considerable number of comprehensive recent monographs and handbook
articles treat the environmental degradation of polymers (e.g. [196-203 D. They
deal in depth with those aspects of environmental degradation which cannot be dis-
cussed any further at this point: thermal degradation, fire and heat stability, chemical
degradation, weathering, aging, moisture sensitivity, effects of electromagnetic and
particle irradiation, cavitation and rain erosion, and biological degradation. For any
detailed information on the above subjects and on methods and agents to protect
and stabilize polymeric materials the reader is referred to the cited comprehensive
references [196-203,207 -209].
It mainly remains to be discussed in this section the nature of the effect which
the simu ltaneou s action of mechanical and environmental parameters may have on
mechanical properties. One of the most interesting phenomena in this field, the
environmental stress crazing (ESC), will be treated in the final chapter.
A characteristic stress-induced degradation process is that of unsaturated rub-
bers in the presence of ozone. The rates of crack initiation and growth, of forma-
tion of free radicals, and of stress relaxation and creep are amplified up to and more
than a thousandfold by the action of ozone [196-197,199,201,204-206]. The
chemical reaction is not completely understood. It is normally assumed that the
237
first steps of ozone degradation of unsaturated polymers follow the "Criegee me-
chanism".
Under the effect of molecular ozone the 1T-bonds of olefines are polarized and
ozone joins the double bonds to form primary ozonides [196, 197a]. Primary ozonides,
-CH - CH-, are unstable, they either isomerize into iso-ozonide, -HC - 0 - CH-,
~ o-if
or polymerize, or decompose into chains terminated by a zwitterion, Rl COO-; and
a carbonyl group, R2 CO, respectively. If a bond attacked by ozone is under stress, de-
composition can occur during the isomerization reation.
The reaction with ozone obviously is a surface reaction leading to the formation
of a surface layer of ozonides and/or further reaction products. This layer grows in
thickness in proportion to the square root of the time of exposure [199]. With in-
creasing layer thickness ozone is gradially prevented from reaching undegraded rub-
ber. The mechanical aspects of ozone cracking are extensively reviewed in the cited
articles [196-197,199,201, 204-206]. It is unanimously reported that the degraded
rubber material (natural rubber, SBR, acrylonitrile-butadiene, cis-polybutadiene) has
a reduced strength and elasticity. Cracks open up and propagate at strains as low as
5 to 12%; it has been estimated [199] that even at the tip of a slowly propagating
crack strains do not exceed 30% (as compared with 700% in an undegraded matrix).
As a consequence of the chemical destruction of the material at the tip of a crack the
mechanical energy expended in crack propagation (material resistance R) is exceed-
ingly small. Employing the Griffith fracture criterion the various authors have ob-
tained values of 0.05 to 0.12 J/m 2 from observation of macroscopic cracks [197a,
199a, 204, 205] or of 0.4 to 0.5 J/m 2 from micromorphological theories [206].
The formation of free radicals in unsaturated rubbers under simultaneous attack
of ozone and stress is of particular interest. The primary steps of the above described
reaction of ozone with unsaturated bonds do not lead to free radicals. In cis-poly-
butadiene (BR), natural rubber (NR), and acrylonitrile-butadiene (ABR), however,
large quantities of peroxy radicals have been observed [206,208]. As one possibility
it was pointed out that these radicals may derive from the ozonides or zwitterions
through unknown secondary steps pOSSibly involving hydrogen abstraction or proton
migration [197, 206, 208]. Another possibility certainly is that radicals are formed
through scission of undegraded rubber molecules and that these primary radicals react
with molecular oxygen. In BR and ABR the concentration of free radicals has shown
the same dependency on strain and ozone concentration as the visual damage, i.e.
the ozone cracking of the surface of the degrading rubber specimen. Notably the
following observations may be mentioned [206, 208]:
no (detectable quantities of) lree radicals are formed below certain threshold
strains (of between 5 and 12%)
for strains beyond the threshold strain the rate of increase in free radical concen-
tration increases initially linearly with strain and ozone concentration (in the range
of 0.08 to 1.6 mg/l)
- increasing strain or ozone concentration lead to a preferential increase in number
rather than size of microcracks.
238
The number of free radicals (up to 8 . 10 14 spins per cm 2 rubber surface) cor-
responds at a given ozone concentration of 2.8 mg/l to the number of ozone molecules
in a zone of surrounding atmosphere of 0.02 cm depth. This number is very much
smaller than the total number of ozone molecules available in the ambient atmos-
phere of the degraded specimens. The linear dependency of radical concentration
on ozone concentration and strain, therefore, means that potential radical sites are
created only after the application of stress and in the presence of ozone molecules.
The appearance of free radicals simultaneously with crack opening certainly indi-
cates the breakage of chains during the process of crack opening. In Section I E of
this chapter an analysis of the energy content of an unsevered hydrocarbon chain
strained elastically up to chain scission had been carried out. This analysis leads to
the estimate that the 2 . 10 14 chains per cm 2 of new fracture surface can only be
broken elastically if more than 91/m 2 were expended in forming that surface. Since
the observed surface work parameters are much smaller one has to conclude that
under ozone attack the breaking and radical forming chains break at small strains.
This interpretation looks identical to that derived from analysis of the behavior of
ozone cracks. Its importance lies in the statement, however, that during crack open-
ing chains (of very low strength) are broken and not only broken chains are separated.
The breakage of ozone attacked chains is a conspicuous example of the syner-
gistic effect of the simultaneous action of mechanical and environmental parameters.
There are many other environmental parameters (e.g. moisture or oxygen content)
which have an accelerating effect in a given situation [196-203]. Of these observa-
tions only a few will be discussed here namely the ones which characterize respec-
tively the chemical aging of stressed rubbers [209c-21 0], the role of moisture in
fatigue of PA 66 and PC [211-212], and the accelerating effect of UV irradiation
on sub micro crack formation and breakdown of highly oriented polymers [74,
213-214].
The oxidative degradation of polymers has been recognized and studied for quite
some time (cf. the extensive review 209a). Most investigations in this field deal with
the effect of degradation on the properties of polymers (e.g. 189,202, 203, 207,
20ge) or with an elucidation of the nature and kinetics of the intervening reactions
[200,201,207,208, 209a-d]. As one of the first Kim [210] also searched for a
possible stress effect on the rate of chain degradation [209d, 210]. She employed a
constant-load technique which permitted to derive the relative number of network
chains, N(t)/N(O), by comparing the Mooney-Rivlin plot, ao/(A.-1/"),2), of an aging
sample with that of a fresh sample. Her interesting results may be summarized as
follows:
The creep rate of polybutadiene rubber films loaded in ambient atmosphere was
about a thousandfold larger than the creep rate under dry nitrogen.
A comparison of the relative potency of the chemical agents present in the ambient
atmosphere showed that the few parts per hundred million of naturally occurring
ozone had a much stronger effect than the oxygen.
In ozone-containing air a behavior comparable to that described above for higher
ozone concentrations had been found. A critical stress of 0.1 MN m -2 had to be
exceeded in order to influence the initial creep rate. The (initial) rates of chain
cleavage dN(t)/N(O)dt were found to be independent of stress within a stress range
239
of 0.1 < 00 < 0.5 MN m- 2 and amounted to 2 . 10-6 S-I. This phenomenon was
explained in the same way as for higher ozone concentrations through formation
and cracking of a degraded surface layer. Microscopical investigations of the sample
surface confirmed such a statement.
- In the absence of ozone the rates of creep and chain scission were greatly reduced.
The effect of oxygen at room temperature was small and hardly any differences
between creep rates in dry nitrogen, dry air without ozone, and ozone-free air of
40% relative humidity had been observed. At slightly higher temperatures (49°C),
however, creep rates in dry air accelerated whereas those in inert atmosphere
(Argon) remained constant.
- From her experiments Kim concludes that in the initial phase of degradation in
ozone-free oxygen stress or strain have no effect on the rate of chain cleavage.
After the chemical degradation had reached a stage, however, where cracks
opened up under stress, the straining of a sample accelerated the failure process
of the specimen through stress concentration and change in cross-linking behav-
ior [210].
The general effect of cross-link density on the elastic modulus of an elastomer
is indicated by Eq. 2.3. In their paper Landel and Fedors [189] consider the in-
fluence of a time-dependent cross-link density on the shape of the stress-strain
curves of silicon, butyl, natural, and fluorinated rubbers. Introducing an additional
shift-factor ax related to the cross-link density, they were able to represent reduced
breaking stresses as a function of reduced time in one common master curve.
Thermal and chemical degradation of cross-linked rubbers was also reviewed by
Murakami [209c] who especially analyzed whether cleavage occurs along main chains
or at cross-links. Depending on the nature of the chain and on that of the cross-links
both mechanisms are observed. Natural rubber, radiation cross-linked to various
degrees, showed at 20% strain and at temperatures of 80 to 130°C random scission
of network chains. An EPT rubber cross-linked by tetramethyl thiuram disulfide
(TMTD) showed degradation at the junction points of network chains. There the S
and S2 linkages seemed to be preferred scission points. In TMTD-cured natural
rubber both types of degradation occur simultaneously. The dependency of the
stress relaxation curves on time, temperature, and initial cross-link density permitted
discrimination between thermal and oxidative degradation. For differently cured
natural rubber (peroxide-, irradiation-, TMTD-, and sulfur-cured) Murakami [209c]
has tabulated the degradation mechanisms.
The accelerating effect of mechanical stresses on chemical stress relaxation
clearly became apparent through the comparison of continuous and intermittent
stress relaxation. The relative stress relaxation, 1-0 (t, )0...)/0 (0, )0...), under continuous
loading was the more pronounced the higher )0.. and always consistently larger than
under intermittent load application. Murakami [209c] also discussed possible increases
of relative stresses due to cross-linking reactions and their partial prevention through
radical acceptors.
With respect to the general effect of an active liquid environment on a polymer
reference to the literature must be made [196-199]. This applies especially to
quantitative data concerning the reduction in strength accompanying the swelling
or plasticization of a material. The penetration of low molecular weight molecules
240
into a polymer generally reduces the intermolecular attraction and facilitates chain
slippage. In the case of hydrogen-bonded materials the possible interaction of the
diffUSing liquid with the hydrogen bonds has to be considered.
In a series of experiments Hearle et al. [211] studied the effect of air, water,
hydrochloric acid, sodium hydroxide, and a laundering detergent on the strength
and fatigue of PA-66 fibers. Their results confirm that the breaking strength of a
PA-66 fiber immediately after immersion in water is 5% smaller than that in air.
Pre-conditioning by immersion for 3 hours causes an additional strength reduction
of 3%. Immersion in aqueous solutions of NaOH and HCI with ph-values larger than
2.5 has an effect similar to that of water. In the range of ph-values between 2 and 0,
however, strength decreases linearly with decreasing ph up to a strength loss of 20%.
Notable is the observation that the pre-conditioning of fibers before making a
fatigue test (by rotation over a wire) does not have such a great effect on the
fatigue life. This may be explained by the fact that the slow growth rates of a
fatigue crack starting from the surface permit the environmental agent to always
penetrate the crack growth region [211].
The partial penetration of a liquid or vapor into a matrix establishes concen-
tration gradients which do lead to a direct mechanical action through non-uni-
form swelling or to indirect actions through the nonhomogeneous relaxation or
distribution of stresses. These actions are even enhanced in the presence of ther-
mal gradients and may cause rapid crack or craze formation. In the case of a slowly
penetrating environment in a well entangled, homogeneous matrix the induced stresses
generally can be accommodated elastically or viscoelastically. For example in poly-
carbonate sheets exposed to artificial weathering no cracking becomes apparent even
after severe cycling of temperature and humidity [212]. Within a relatively short
period of 30 to 32 months of outdoor weathering, however, a network of surface
microcracks developed on the side exposed to the solar radiation. By comparison
with artificial UV irradiation the authors [212] were able to show that the photo-
chemical degradation of the surface layer introduces defects and lowers the strength
of the polymeric material to such an extent that physically induced non-homoge-
neous stresses cause micro cracking rather than being accommodated. The effect of
liquid environment on crack or craze formation will be further discussed in 9 IID.
Photo-chemical degradation is probably the most important factor of environ-
ment. The cited monographs [196-203, 207-209] treat in detail the primary pro-
cesses of photon ab sorption, electron excitation, energy transfer through excitons,
luminescence, phosophorescence, and raditionless transitions, chain scission and free
radical formation, secondary reactions, stabilization and protection, and the effect
of irradiation on the mechanical properties. Here only one point can be touched:
the accelerating effect of UV radiation on micro crack formation in loaded polymers
[74,213-214].
As indicated by Tables 4.4 and 4.5 photon energy in the spectral range of
250 < A < 360 nm is larger than the dissociation energy of C-C bonds. The high
energy tail of solar radiation extends to about A = 300 nm. Chain scission in UVand
solar light is, therefore, energetically possible. In the near UV in unprotected glassy
polymers quantum yields of the order of 1 . 10- 3 to 60 . 10- 3 are observed [209b].
241
The principle mechanisms of the synergistic action of load and UV radiation

may be discussed using the few data available. The simultaneous application of ten-
sile load and UV irradiation to oriented polymers clearly accelerated the formation
of free radicals and/or of micro and macro cracks in PA 6 [213-214] and in natural
silk, cotton, and "triacetate" fibers [213]. No effect was observed in PMMA [213].
The experiments on cotton and triacetate fibers revealed that at low tensile stresses
(uo < 70 MN m -2) the UV irradiation reduced the lifetimes of the fibers by more
than 4 orders of magnitude. Under these conditions the absence or presence of
oxygen was of some, but minor, importance since the irradiation in vacuum led to
only slightly longer lifetimes than the irradiation in air. In a range of stresses with
70 < Uo < 220 MN m -2 no oxygen effect on the lifetime of triacetate fibers became
apparent. In this stress range the irradiation effect decreased with increasing uo. For
Uo > 220 MN m- 2 the lifetime depended only on stress and not on the environ-
mental factors UV irradiation and oxygen content. With cotton fibers a somewhat
similar behavior was observed although the upper stress limit was smaller and de-
pended on absence or presence of air [213]. The described behavior points to the
existence of three failure mechanisms which are occurring simultaneously and at
different rates: oxydation, UV degradation, and creep. An effect of oxydation was
observed for the acetate fibers only at lifetimes tb > 5 . 10 3 s and in the presence of
UV radiation. At shorter lifetimes, 100 < tb < 5 . 103 s, failure was essentially due
to irradiation. At very short lifetimes (and higher stresses) oxydation and UV degrada-
tion were irrelevant if compared to the stress effect.
The determination by X-ray scattering of the number of microcracks in a PA-6
fiber stressed in air at 128 MN m- 2 gave the notable result [214] that the rate of
micro crack accumulation increased almost instantaneously (from 5 . 10 16 m- 3 S-1
to 110· 10 16 m- 3 s-l) with the application of UV irradiation. The rate decreased
to its initial value in the same abrupt and repeatable manner when UV was switched
off after 10 4 s. Irradiation of an unstressed sample did not cause any micro crack
formation and did not influence subsequent formation rates. In a previous study
[213] it had been shown that UV irradiation of stressed PA-6 and natural silk fibers
in a helium atmosphere increased the accumulation of free radicals. In that case the
rate of radical accumulation at 200 < Uo < 600 MN m- 2 decreased with the length
of the time of irradiation and reached a steady state concentration NCR) after
5 . 103 s. In PA 6 at a stress of 600 MN m -2 the concentration NCR) was of the
order of 1024 m- 3 ; this is about the limiting concentration observed in purely stress-
induced chain scission also.
From these results the following conclusions can be derived. The photodegrada-
tion of unstressed PA 6 in air most probably is a random oxidative chain scission
process [215]. Based on a study of the formed ESR spectra Heuvel et al. [216] con-
cluded that amide bonds in the amorphous regions are broken leading to R-CO and
NH-R' free radicals. Subramanian et al. [215] propose formation ofhydroperox-
ides and keto-intermediates at the a-CH2 group and a subsequent hydrolytic break-
down at the RCO-NHCOR' bond. It must be assumed that in PA-6 in air under the
combined action of UV and stress all chain segments in the amorphous regions are
possible fracture sites and not only the most highly stressed ones. The rates of chain
degradation and cross-linking may depend on chain stress through the mechanisms
242
Kererences lOr e-napler 1\
discussed below. In the absence of oxygen, UV irradiation accelerated the breakage

of chains but did show a saturation at a concentration N(R). This result can be
explained if one assumes that in irradiated and stressed PA 6 preferentially those
chains are broken and are prevented from recombination which are most highly
stressed.
Several mechanisms are conceivable through which the UV degradation of highly
extended and stressed chains would be facilitated. In the first place the decrease of
bond dissociation energies through electronic excitation and tensile stress has to be
mentioned (cf.4UB). In the case of ionized hydrocarbon chains, bond dissociation
energies are as small as 100 kJ/mol (Table 4.6). As outlined in Chapter 7 bonds of
such energy require rather small stresses for breakage (about 114 to 115 of the
strength of a non-ionized chain). In addition to the reduction of the dissociation
energy an increase in local thermal excitation - due to energy transfer or dissipa-
tion - may have an effect. Chain orientation and segment conformation also
influence the effective stability of a chain. Very little is being said in the literature
about the relation between molecular order and UV resistance. Reinisch et al. [217]
investigated the influence of sample orientation on photo-chemical degradation.
They found that highly drawn PA-6 fibers (A. = 3.9) had an even higher degradation
resistance than undrawn material. Krlissmann et al. [218] indicate that the reactivity
(in photo oxidation) of N-vicinal methylen groups in trans conformation is 60 times
that of those in gauche conformation. These data [213-218] were obtained for
quite different specimens and experimental conditions. They do not yet permit a
quantitative estimate as to the importance and interaction of the above mechanisms.
Electronic excitation, ionization, radical formation, oxidation, and cross-linking
are also the principal processes occurring in polymer solids subjected to nuclear
radiation (a, (3, 'Y, nucleons). In view of the fact that the molecular mobility influences
the kinetics of degradation and cross-linking a synergistic stress effect is conceivable
but not yet proven. The current investigations aim at an understanding of the inter-
relation between irradiation characteristics (dosis and dosis rate), network structure,
and macroscopic properties after irradiation [198, 200, 219].
1. S. N. Zhurkov, A. Ya. Savostin, E. E. Tomashevskii: Dokl. Akad. Nauk SSSR 159,

303 (1964). Soviet Phys. "Doklady" (Eng!. Trans!.) 9,986 (1964).
2. H. H. Kausch, 1. Becht: Rheo!. Acta 9, 137 (1970).
3. V. R. Regel, A. I. Slutsker, E. E. Tomashevskii: "The Kinetic Nature of the Strength of
Solids" (in Russian), Moscow: Isdat. Nauka, 1974.
4. S. N. Zhurkov, V. I. Vettegren', 1.1. Novak, K. N. Kashintseva: Doklady Akad. Nauk
SSSR 176/3,623-626 (1967). Soviet Phys. "Doklady" (Eng!. Trans!.) 176/3,708-711
(1967).
5. S. N. Zhurkov, V. I. Vettegren', V. E. Korsukov, 1.1. Novak: Fizika Tverdogo Tela
11/2,290-295 (1969). Soviet Phys. Solid State 11/2,233-237 (1969).
243
6. S. N. Zhurkow, V. I. Vettegren', V. E. Korsukov, 1.1. Novak: Fracture 1969, paper IV/47.

7. A. I. Gubanov: Mekh. Polimerov 3/4,608-614 (1967).
8. A.1. Gubanov: Mekh. Polimerov 3/5,771-776 (1967).
9. A. I. Gubanov: Mekh. Polimerov 4/4,586-594 (1968).
10. V. E. Korsukov, V. I. Vettergren', I. I. Novak: Mekh. Polimerov 6/1,167 -170 (1970).
Polym. Mech. (USA) 6/1,156-159 (1972).
11. V. A. Kosobukin: Mekh. Polimerov 6/6, 971-978 (1970) Polym. Mech. (USA) 6/6,
846-852 (1970).
12. V. A. Kosobukin: Fizika Tverdogo Tela 14/9, 2595-2602 (1972) Soviet Phys. Solid State
14/9, 2246-2251 (1973).
13. V. A. Kosobukin: Mekh. Polimerov8/1, 3-11 (1972) Polym. Mech. (USA) 8/1,1-7 (1972).
14. V. A. Kosobukin: Opt. Spektrosk. 34/2, 273-277 (1973). Opt. & Spectrosc. (USA) 34/2,
154-156 (1973).
15. V.1. Vettegren', 1.1. Novak: Fizika Tverdogo Tela 15,1417-1422 (1973). SOY. Phys.
Solid State 15/5, 957-960 (1973).
16. V. I. Vettegren', 1.1. Novak, K. J. Friedland: Int. J. Fracture 11,789-801 (1975).
17. S. N. Zhurkov, V. A. Zakrevskii, V. E. Korsukov, V. S. Kuksenko: Fizika Tverdogo
Tela 13/7,2004-2013 (1971). Soviet Phys. Solid State 13/7,1680-1688 (1972).
18. S. N. Zhurkov, V. E. Korsukov: Fizika Tverdogo Tela 15,2071-2080 (1973). Soviet
Phys. Solid State 15/7, 1379-1384 (1974)
S. N. Zhurkov, V. E. Korsukov: J. Plym. Sic. (Polymer Phys. Ed.) 12/385-398 (1974).
19. A. Ya. Savostin, E. E. Tomashevskii: Fizika Tverdogo Tela 12/10, 2857 -2864 (1970).
Soviet Phys. Solid State 12/10, 2307-2311 (1971).
20. V. A. Zakrevskii, V. Yeo Korsukov: Vysokomol. soyed. B13, 105 (1971). Vysokomol.
soyed. A14/4, 955-961 (1972). Polym. Sci. USSR 14, 1064-1071 (1972).
21. B. Ya. Levin, A. V. Savitskii, A. Ya. Savostin, E. Yeo Tomashevskii: Vysokomol. soyed.
A13/4, 941-947 (1971). Polymer Sci. USSR 13,1061-1068 (1971).
22. S. I. Veliev, V.I. Vettegren', 1.1. Novak: Mekh. Polimerov 6/3, 433-436 (1970).
Polym. Mech. (USA) 6/3,369-372 (1970).
23. u. G. Gafurov: Mekh. Polimerov 7/4, 649-653 (1971). Polym. Mech. (USA) 7/4,578-581
(1971).
24. V.1. Vettegren', V. Yeo Korsukov, 1.1. Novak: Plaste und Kautschuk 19/2, 86-88 (1972).
25. V. E. Korsukov, V.1. Vettegren', 1.1. Novak, A. Chmel: Mekh. Polimerov 4,621-625
(1972). Polym. Mech. (USA) 4,536-539 (1972).
26. V. Yeo Korsukov, V. I. Vettegren', I. 1. Novak, L. P. Zaitseva: Vysokomol. soyed
A16/7, 1538-1542 (1974). Polymer Sci. USSR 16, 1781-1785 (1974). E. E. Tomashevskii,
V. A. Zakrevskii, I. I. Novak, V. E. Korsukov, V. R. Regel, O. F. Pozdnyakov, A. I. Slutsker,
V. S. Kuksenko: Int. J. of Fracture 11/5,803-815 (1975).
27. S. N. Zhurkov, V. S. Kuksenko, A.1. Slutsker: Fizika Tverdogo Tela 11/2,296-307
(1969). Soviet Phys. Solid State 11/2, 238- 246 (1969).
28. V. S. Kuksenko, A. I. Slutsker: Mekh. Polimerov 6/1, 43-47 (1970). Polym. Mech. (USA)
6/1,36-40 (1970). V. M. Knopov, V. S. Kuksenko, A. I. Slutsker: Mekh. Polimerov 6/3,
387-392 (1970). Polym. Mech. (USA) 6/3,329-333 (1970).
29. V. S. Kuksenko, V. S. Ryskin, V. I. Betekhtin, A. I. Slutsker: Int. J. of Fracture 11/5,
829-840 (1975).
30. A. V. Amelin, O. F. Pozdnyakov, V. R. Regel: Mekh. Polimerov 4/3, 467-473 (1968).
Polym. Mech. (USA) 4/3,376-379 (1968). A. V. Amelin, O. F. Pozdnyakov, V. R. Regel,
T. P. Sanfirova: Fizika Tverdogo Tela 12/9, 2528-2534 (1970). Soviet Phys. Solid
State 12/9, 2034-2038 (1971). A. V. Amelin, Yu. A. Glagoleva,A. O. Podol'skii, O. F.
Pozdnyakov, V. R. Regel, T. P. Sanfirova: Fizika Tverdogo Tela 13/9,2726-2731
(1971). Soviet Phys. Solid State 13/9,2279-2283 (1972).
31. E. E. Tomashevskii: Fizika Tverdogo Tela 12/11, 3202-3207 (1971). Soviet Phys. Solid
State 12/11,2588-2592 (1971). Yu. K. Godovskii, V. S. Papkov, A.1. Slutsker, E. E.
Tomashevskii, and G. 1. Slonimskii: Fizika Tverdogo Tela 13/8,2289-2295 (1972).
Soviet Phys. Solid State 13/8,1918-1923 (1972).
244
32. E. E. Tomashevskii, E. A. Egorov, A. Va. Savostin: Int. J. Fracture 11/5, 817-827 (1975).
33. V. R. Regel, O. F. Pozdnyakov: Plaste u. Kautschuk 19/2, 99-100 (1972).
34. D. K. Roylance, K. L. DeVries: Polymer Letters 9,443-447 (1971).
35. R. P. Wool: PhD Thesis, Univers. of Utah, Dept. of Materials Science and Engng., Salt
Lake City, Utah 1974. R. P. Wool: J. Polymer Sci. 12, 1575-1586 (1974).
36. R. P. Wool: J. Polymer Sci. 13, 1795-1808 (1975).
37. R. P. Wool: J. Polymer Sci. 14, 1921-1929 (1976).
38. R. P. Wool, H. H. Kausch: Unpublished results, Salt Lake City 1974.
39. R. P. Wool, W. O. Statton: In press.
40. K. K. R. Mocherla. W. O. Statton: Symposium for High Polymer Physics (1975), 1-13.
K. K. R. Mocherla: PhD Thesis, Univers. of Utah, Dept. of Materials Science and Engng.,
Salt Lake City, Utah 1976.
41. H. Tadokoro et al.: J. chern. Phys. 42,4 (1965).
42. T. Miyazawa: J. Polymer Sci. C2, 59 (1974).
43. J. P. Luongo: J. Appl. Polymer Sci. III/9, 302-309 (1960).
44. J. Becht: Dissertation, Technische Hochschule Darmstadt, 1970.
45. J. Becht, H. Fischer: Kolloid-Z. u. Z. Polymere 240, 766-774 (1970).
46. F. Sziics, J. Becht, H. Fischer: Europ. Polymer J. 7, 173-179 (1971).
47. J. Becht, H. Fischer: Angew. Makromol. Chern. 18, 81-91 (1971).
48. H. H. Kausch, J. Becht: Deformation and Fracture of High Polymers, 317 - 3 33. J. A.
Hassell, R. I. Jaffee, Plenum Press 1974.
49. K. L. DeVries, B. A. Lloyd, M. L. Williams: J. Appl. Physics 42/12,4644-4653 (1971).
50. B. A. Lloyd, K. L. DeVries, M. L. Williams: J. Appl. Polymer Sci.1O/A-2, 1415-1445 (1972).
51. D. Klinkenberg: Dissertation, Technische Hochschule, Darmstadt, 1978.
52. T. Nagamura, K. Fukitani, M. Takayanagi: J. Polymer Sci. (Polymer Phys. Ed.) 13,
1515-1532 (1975).
53. D. C. Prevorsek: J. Polymer Sci. A-2, 4,63-88 (1966).
54. H. H. Kausch, K. L. DeVries: Intern. 1. of Fracture 11/5, 727 -759 (1975).
55. H. H. Kausch, J. Becht: Kolloid-Z. u. Z. Polymere 250,1048-1065 (1972).
56. W. O. Statton: J. Polymer Sci. C32, 219 (1971).
57. K. L. DeVries: J. Polymer Sci. C32, 325 (1971).
58. A. Peterlin: Internat. J. of Fracture 11/5,761-780 (1975).
59. F. H. MUller, A. Engelter: Kolloid-Z. u. Z. Polymere 149/2-3,126-127 (1956).
60. F. H. MUller, A. Engelter: Rheolog. Acta 1,39-53 (1958).
61. A. Engelter, F. H. MUller: Kolloid-Z. u. Z. Polymere 157/2,89-111 (1958).
62. F. H. MUller: Kunststoffe 49/2,67-71 (1959).
63. F. H. MUller, N. Weimann: J. Polymer Sci. C6, 117-124 (1964).
64. J. Stiilting, F. H. MUller: Kolloid-Z. u. Z. Polymere 238/1-2,459-470 (1970). J. Stiilting,
F. H. MUller: Kolloid-Z. u. Z. Polymere 240,792-806 (1970).
65. W. Dick, F. H. MUller: Kolloid-Z. u. Z. Polymere 172/1,1-18 (1960).
66. W. Dick, A. Engelter, F. H. MUller: Rheo!. Acta 1/4,6,506-510 (1961).
67. F. H. MUller: J. Polymer Sci. C20, 61-76 (1967).
68. F. H. MUller, A. Engelter: Kolloid-Z. u. Z. Polymere 171 /2, 152-153 (1960).
69. F. H. MUller, J. Stiilting: Kolloid-Z. u. Z. Polymere 240,790-791 (1970).
70. V. I. Vettegren', A. E. Chmel: Europ. Polymer J. 12,853-858 (1976).
71. A. C. Lunn, I. V. Yannas: 1. Polymer Sci. (Polymer Phys. Ed.) 10, 2189-2208 (1972).
72. W. F. Busse, E. T. Lessig, D. L. Loughborough, L. Larrick: J. App!. Physics 13, 715 - 724
(1942).
73. W. J. Lyons: Text. Res. J. 28,127 (1958).
74. V. R. Regel, A. M. Leksovskii: Int. J. Fracture Mechanics 3/2,99-109 (1967). V. R.
Regel, A. M. Leksovskii: Polymer Mechanics 5,58-78 (1969) Mekhanika Polimerov 5,
70-96 (1969). V. R. Regel: Mekhanika Polimerov, 7,98-112 (1971).
75. V. P. Tamush: Polymer Mechanics 5/1,79-87 (1969). Mekhanika Polimerov 5/1, 97-107
(1969).
76. J. W. S. Hearle, E. A. Vaughn: Rheologica Acta 9/1, 76-91 (1970).
245
77. A. R. Bunsell, J. W. S. Hearle: J. Materials Sci. 6, 1303-1311 (1971).

78. D. C. Prevorsek, W. J. Lyons: Rubber Chemistry and Technol. 44/1,271-293 (1971).
79. A. R. Bunsell, J. W. S. Hearle: J. Appl. Polymer Sci. 18, 267-291 (1974).
80. V. Regel, V. P. Tamush: Mekhanika Polimerov 3,458-478 (1977).
81. J. W. S. Hearle, B. S. Wong: J. Text. Inst. 68/3, 89-94 (1977).
82. S. F. Calil, J. W. S. Hearle: Fracture 1977, Vol. 2, ICF4, Waterloo, Canada, June 1977.
83. H. Hendus, E. Penzel: Chemiefasern/Textilindustrie (1976/6), 527.
84. Courtesy of H. Hendus, Ludwigshafen 1976.
85. J. W. S. Hearle: Proc. Textile Inst./Inst. Text. de France Conf. Paris (1975), 60-75.
86. Courtesy to J. W. S. Hearle, Manchester 1977.
87. J. W. S. Hearle: "An Atlas of Fibre Fracture". Text. Mfr. (1972), 99, Jan., Feb., 14; March,
12; May, 20; Aug., 40; Sept., 16; Oct., 40; Nov., 12; Dec., 36; (1973), lOa, Jan., 24;
March, 24; April, 34; May, 54; June, 44.
88. P. I. Vincent: Impact tests and service performance of thermoplastics, Plastics Inst., London,
1971.
89. C. B. Bucknall, K. V. Gotham, P. I. Vincent in: Polymer science, A. D. Jenkins (ed.), North-
Holland PUbl., Chapt. 10, 1972.
90. H. Oberst: Kunststoffe 52/1,4-11 (1962).
91. W. Retting: Kolloid-Z. u. Z. Polymere 210/1,54-63 (1966).
92. M. H. Litt, A. V. Tobolsky: J. Macromol. Sci.-Phys., B1 (3),433-443 (1976).
93. J. Heijboer: J. Polym. Sci. (C) 16,3755 (1968).
94. R. F. Boyer: Polym. Engng. Sci. 8, 161 (1968).
95. J. A. Sauer: J. Polymer Sci. C32, 110-116 (1971).
97. W. Retting: Materialpriifung8/2, 55-60 (1966).
98. H. Grimminger, G. Koch, J. Penzkofer, E. P. Petermann, W. Retting, J. Steinig: Kunst-
stoffe 59/6,375-381 (1969).
99. K. Fujioka: J. Appl. Polymer Sci. 13, 1421-1434 (1969).
100. H. H. Racke, T. Fett: Kunststoffe 64/9, 481-487 (1974).
101. G. Hoff, G. Langbein: Kunststoffe 56, 2-6 (1966).
102. H. Oberst: Kunststoffe 59/4, 232- 240 (1969).
103. W. B. Hillig: Impact Response Characteristics of Polymeric Materials, Techn. Inform. Series,
General Electric Co., Report No. 76CRD271, New York 1976.
104. H. Saechtling: Kunststoff-Taschenbuch, 20. Ausg., 910-914, 516, 518, Miinchen: Carl
Hanser 1977.
105. F. Ramsteiner: Kunststoffe 67/9, 517-522 (1977).
106. a) E. Gaube, H. H. Kausch: Fracture Theories in Industrial Use of Thermoplastics and
Glassfiber Reinforced Plastics, Intern. Conf. Fracture, Munich 1973, Vol. I, PI VI-311.
106. b) E. Gaube, H. H. Kausch: Kunststoffe 63/6,391-397 (1973).
107. Courtesy to E. Gaube, Frankfurt-Hoechst.
108. W. Retting: Angew. Makromol. Chern. 58/59, 133-174 (1977).
109. T. T. Jones: Effect of Molecular Orientation on the Mechanical Properties of Polystyrene,
Rep. IUP AC Working Party, IUPAC Internat. Symposium on Macromolecules, 10-14 Sept.
1973, 41-57.
110. W. Retting: The Effect of Molecular Orientation on the Mechanical Properties of Rubber-
Modified-Polystyrene, to be published in "Pure and Applied Chemistry".
111. H. Niklas, H. H. Kausch: Kunststoffe 53/12, 886-891 (1963).
112. I. V. Yannas: J. Polymer Sci. 9,163-190 (1974).
113. D. W. Hadley, I. M. Ward: Reports on Progress in Physics 38/10,1143-1215 (1975).
114. S. S. Stern stein in: Treatise on Materials Science and Technology, J. M. Schultz, ed.,
Vol. 10, New York: Academic Press 1977,567-569.
115. E. N. C. da Andrade: Proc. Royal. Soc. (London) A 84,1 (1910).
116. Taprogge, R.: Kunststoff-Rundschau 15/5-12, 3-50 (1968).
117. E. Gaube, G. Diedrich, W. Miiller: Kunststoffe 66/1,2-8 (1976).
118. P. Stockmayer: Stuttgarter Kunststoff-Kolloquium 1977; Kunststoffe 67,470 (1977).
246
References tor Chapter l:I
119. A. V. Savitskii, V. A. Mal'chevski, T. P. Sanf"lrova, L. P. Zosin: Polymer Sci. USSR 16,

2470-2477 (1974). Vysokomol. soyed. A16/9, 2130-2135 (1974).
120. S. N. Zhurkov, B. N. Narzulayev: J. Techn. Phys. 23,677 (1953).
121. D. F. Kagan, A. M. Knebel'man, L. A. Kantor: Vysokomol. soyed A14/5, 1207-1214
(1972).
122. M. J. Mindel, N. Brown: J. Materials Sci. 9, 1661-1669 (1974).
123. L. C. E. Struik: Polym. Eng. and Sci. 17/3, (1977). L. c. E. Struik: Physical Aging in Amor-
phous Polymers and Other Materials Amsterdam/New York: Elsevier Scientific Publishing
Co. 1978.
124. S. Matsuoka, H. E. Bair: J. Appl. Physics 48/10, 4058-4062 (1977). S. Matsuoka, H. E.
Bair, S. S. Bearder, H. E. Kern, J. T. Ryan: Analysis of Nonlinear Stress Relaxation in
Polymeric Glasses, Bell Laboratories, Murray Hill, New Jersey 07974.
125. J. F. Jansson, B. Terselius: IUPAC Symposium on Long-Term Properties of Polymers,
Stockholm, Sweden 30. 8. - 1. 9. 1976.
126. E. H. Andrews: Fatigue in Polymers, Testing of Polymers, Vol. IV, W. Brown, ed., New York:
Interscience 1969,237.
127. J. A. Manson, R. W. Hertzberg: CRC Critical Reviews in Macromol Sci. 433-500, August
1973.
128. R. W. Hertzberg: Deformation and Fracture Mechanics of Engineering Materials, New York:
Wiley 1976.
129. K. Oberbach: Kunststoffe 63,35-41 (1973).
130. K. Oberbach: Kunststoff-Kennwerte ftir Konstrukteure, Miinchen: Carl Hanser Verlag
1975,87-96.
131. J. M. Schultz: Treatise on Materials Science and Technology, Vol. 10, Part B. New York:
Academic Press 1973, 80-107.
132. G. Jacoby in: Neuzeitliche Verfahren der Werkstoffpriifung, Diisseldorf: Verlag Stahleisen
1973,80-107.
133. C. E. Feltner, M. R. Mitchell: ASTM STP 465, Amer. Soc. Test. Mat., 27 (1969).
134. K. Oberbach, G. Heese: Materialpriifung 14/6, 173-178 (1972).
135. Crawford, R. J., Benham, P. P.: J. Mech. Engng. Sci. (GB) 16/3,178-179 (1974).
136. M. E. Graf, M. Ya. Filatov: Probl. Prochn. (USSR) 7/9, 102-106 (1975).
137. R. J. Crawford, P. P. Benham: J. Materials Sci. 9, 1297-1304 (1974).
138. V. Bouda, A. J. Staverman: J. Polymer Sci. Polymer Sci. (polymer Phs. Ed.) 14,
2313-2323 (1976).
139. V. Zilvar: J. Macromol. Sci.-Phys. B5/2, 273-284 (1971).
140. V. Zilvar: Plastics and Polymers, 328-332, October 1971.
141. E. Alf: Untersuchungen zum Verhalten ausgewiihlter Kunststoffe unter schwingender
Beanspruchung (Dissertation), Technische Hochschule Aachen 1972.
142. S. Rabinowitz, A. R. Krause, P. Beardmore: Materials Sci. 8, 11-22 (1973).
143. M. Schrager: J. of Polymer Sci. Part A-2 8,1999-2014 (1970).
144. S. Sikka: PhD Thesis, Univers. of Utah, Salt Lake City, Utah.
145. J. H. Wendorff: Personal communication, D. Kunststoff-Inst., Darmstadt (1977).
146. T. Nagamura, N. Kusumoto, M. Takayanagi: J. Polymer Sci. (Polymer Phys. Ed.) 11,
2357-2369 (1973).
147. o. W. Briiller, N. Brand: Kunststoffe 67/9, 527 (1977).
148. L. Konopasek, J. W. S. Hearle: J. Appl. Polymer Sci. 21, 2791-2815 (1977).
149. H. H. Kausch: Materialpriifung 20, 22-26 (1978)
150. K. V. Gotham: Plastics and Polymers 40,59-64 (1972).
151. N. T. Smotrin, V. M. Chebanov: Mekhanika Polimerov 6,453-467 (1970).
152. A. N. Machyulis, M. I. Pugina, A. A. Zhechyus, V. K. Kuchinskas, A. P. Stasyunas: Mek-
hanika Polimerov 2/1,60-66 (1966). V. K. Kuchinskas, A. N. Machyulis: Polymer Me-
chanics 4, 538-543 (1968).
153. J. A. Sauer, A. D. McMaster, D. R. Morrow: J. Macromol. Sci.-Phys. BI2(4), 535-562
(1976). J. A. Sauer, E. Foden, D. R. Morrow: Polymer Eng. and Sci. 17/4, 246-250 (1977).
154. J. C. Bauwens, C. Bauwens-Crowet, G. Homes: J. Appl. Polymer Sci. 7, 1745-1754 (1969).
247
155. J. C. Bauwens: J. App\. Polymer Sci. 8,893-901 (1970).

156. I. M. Ward: J. Materials Sci. 6, 1397-1417 (1971).
157. P. B. Bowden, J. A. Jukes: J. Materials Sci. 7, 52-63 (1972).
158. C. Bauwens-Crowet, J. C. Bauwens, G. Homes: J. Materials Sci. 7, 176-183 (1972).
159. R. Raghava, R. M. Caddell, G. S. Y. Yeh: J. Materials Sci. 8,225-232 (1973).
160. J. A. Sauer, K. D. Pae: Coli. and Polymer Sci. 252,680-695 (1974).
161. A. S. Argon, M. I. Bessonov: Philosophical Mag. 35/4, 917-933 (1977).
162. A. S. Argon, M. I. Bessonov: Polymer Engng. Sci. 17/3, (1977).
163. R. P. Kambour, R. E. Robertson: The Mechanical Properties of Plastics, in Polymer Science,
Chapter 11, Jenkins Ed., Amsterdam, London: North-Holland, 1972,687-822.
164. T. E. Brady, G. S. Y. Yeh: J. App\. Physics 42/12, 4622-4630 (1971).
165. K. P. Grosskurth: Gummi/Asbest/Kunststoffe 25/12,1159-1164 (1972).
166. G. P. Andrianova, V. A. Kargin: Polymer Sci. USSR 12/1,1-8 (1970). Vysokomo\. Soyed.
A12/1, 3-9 (1970).
167. G. P. Andrianova, A. S. Kechekyan, V. A. Kargin: J. Polymer Sci. A-2/9, 1919-1933 (1971).
168. G. I. Barenblatt in: Deformation and Fracture of High Polymers, Kausch, Hassell, Jaffee
Eds., New York: Plenum Press, 1972,91-111.
169. B. Zaks, M. L. Lebedinskaya, V. N. Chaldize: Vysokomo\. Soyed. A12/12, 2669-2679
(1970).
170. L. A. Davis, C. A. Pampillo: J. App\. Physics 42/12, 4659-4666 (1971).
171. L. A. Davis, C. A. Pampillo: J. App\. Physics 43/11,4285-4293 (1972).
172. J. C. M. Li, C. A. Pampillo, L. A. Davis in: Deformation and Fracture of High Polymers,
Kausch, Hassell, Jaffee Eds., New York: Plenum Press, 1972,239-258.
173. L. A. Davis, R. H. Baughman, C. A. Pampillo: J. Polymer Sci.: Polymer Phys., Ed. 11,
2441-2451 (1973).
174. V. A. Lishnevskii: Dokl. Akad. Nauk SSSR, 182/3, 596-599 (1968).
175. V. A. Lishnevskii: Vysokomol. Soedin., Ser. B 11/1,44-49 (1969).
176. P. Matthies, J. Schlag, E. Schwartz: Angew. Chern. 77/7,323-327 (1965).
177. T. C. Chiang, J. P. Sibilia: J. Polymer Sci.: Polymer Phys. Ed. 10, 2249-2257 (1972).
178. U. G. Gafurov: Vysokomo\. Soyed. A14/4, 873-880 (1972).
179. A. K. Sengupta, R. K. Singh, A. Majumdar: Textile Res. J., 155-163, March (1973).
180. N. J. Capiati, R. S. Porter: J. Polymer Sci. (Polymer-Phys. Ed.) 13, 1177-1186 (1975).
D. M. Bigg: Polymer Engineering Sci. 16/11, 725-734 (1976).
181. G. Capaccio, I. M. Ward: Polymer 18/9, 967-968 (1977).
182. Seminar on ultra-high modulus polymers; La Chimica e l'Industria 59, Ottobre 1977,728-735.
183. L. Mullins: Effects of Fillers in Rubber in: The Chemistry and Physics of Rubber-Like
Substances, Chapter 11, Bateman Ed., New York: Wiley, 1963, 301-328.
184. T. L. Smith: Pure and App\. Chern. 23,235-253 (1970).
185. N. Sekhar, B. M. E. van der Hoff: J. Appl. Polymer Sci. 15,169-182 (1971).
186. F. R. Eirich in Mechanical Behavior of Materials Vol. III, Kyoto: Society of Materials Sci.,
1972,405-418.
187. T. L. Smith, W. H. Chu: J. Polymer Sci. A-2/10/1, 133-150 (1972).
188. F. P. Baldwin, G. Ver Strate: Rubber Chemistry and Techn. 45/3,709-881 (1972).
189. R. F. Landel, R. F. Fedors in: Deformation and Fracture of High Polymers, Kausch, Hassell,
Jaffee Eds., New York: Plenum Press, 1972, 131-148.
190. R. J. Morgan: J. Polymer Sci.: Polymer Phys. Ed. 11, 1271-1284 (1973).
191. T. L. Smith in: Treatise on Materials Science and Technology, J. M. Schultz, Ed. Vol. 10,
Part A New York: Academic Press, 1973, 369.
192. R. F. Fedors, R. F. Landel: J. Polymer Sci. (Polymer-Phys. Ed.) 13, 419-429 (1975).
193. R. F. Fedors: J. App\. Polymer Sci. 19, 787-790 (1975).
194. R. F. Fedors, R. F. Landel: J. Appl. Polymer Sci. 19, 2709-2715 (1975).
195. R. F. Fedors: The Stereo Rubbers, W. M. Saltman Ed., New York: John Wiley, 1977.
679-804.
196. Encyclopedia of Polymer Science and Technology, Mark Ed., New York: Interscience, 1965.
a. W. L. Cox, Vol. 2, 197. b. N. Grassie, Yo\. 4, 702.
248
197. G. M. Bartenev, Y. S. Zuyev: Strength and Failure of Visco-Elastic Materials, Oxford:

Pergamon Press 1968.
a. pg. 274.
198. D. V. Rosato, R. T. Schwartz: Environmental Effects on Polymeric Materials, Vol. I + II,
New York: Interscience 1968.
199. E. H. Andrews: Fracture in Polymers, Edinburgh: Oliver and Boyd 1968.
a. pg. 168.
200. A. Charlesby, Radiation Effects in Polymers; in: Polymer Science Vol. 1, Chapter 23,
Jenkins Ed., Amsterdam London: North-Holland, 1972, 1543-1559.
201. Polymer Stabilization, W. L. Hawkins Ed. New York: Wiley-Inter science, 1971.
202. Natiirl. u. kiinstl. Alterung, Kunststoffe, Fortschrittsberichte, Vol. 3, Miinchen 1976: Carl
Hanser.
203. F. H. Winslow: Environmental Degradation, in Treatise on Materials Science and Tech-
nology, J. M. Schultz, ed., Vol. 10, Part B, New York: Academic Press, 1977,741-776.
Developments in Polymer Degradation, N. Grassie, ed., Vol. 1, London: Appl. Science
Publ. 1977.
204. E. H. Andrews: J. Appl. Polymer ScL10, 47 (1966).
205. G. Salomon, F. van Bloois: J. Appl. Polymer Sci. 8,1991 (1964).
206. K. L. DeVries, E. R. Simonson, M. L. Williams: J. Macromol. Sci.-Phys. B4/3, 671 (1970).
K. L. DeVries, E. R. Simonson, M. L. Williams: J. Appl. Polymer Sci. 14, 3049 (1970).
207. Degradation and Stabilization of Polyolefins: B. Sedlacek, C. G. Overberger, H. F. Mark,
T. G. Fox Eds: Polymer Symposia 57 (1976).
208. B. Ranby, J. F. Rabek: ESR Spectroscopy in Polymer Research, Berlin-Heidelberg-New York:
Springer 1977.
209. H. H. G. Jellinek Ed.: Aspects of Degradation and Stabilization of Polymers: Amsterdam-
Oxford: Elsevier 1978.
a) Y. Kamiya, E. Niki: Oxidative Degradation, p. 79-148.
b) W. Schnabel, J. Kiwi: Photodegradation, p. 149-246.
c) I. Mita, Effect of Structure on Degradation and Stability of Polymers, p. 247 -291.
d) K. Murakami: Mechanical Degradation, p. 296-392.
e) H. Kambe: The Effect of Degradation on Mechanical Properties of Polymers, p. 393-430.
210. C. S. Kim: Rubber Chemistry and Technology 42/4,1095-1121 (1969)
211. J. W. S. Hearle, B. S. Wong: J. of the Textile Institute 63/4,127 -132 (1977).
212. A. Blaga, R. S. Yamasaki: J. Materials ScLl1, 1513-1520 (1976).
213. G. G. Samoilov, E. E. Tomashevskii: Fizika Tverdogo Tela 10/4,1094-1097 (1968).
Soviet Physics-Solid State 10/4, 866-869 (1968). T. B. Boboev, V. R. Regel': Mekh.
Polimerov 5,929-931 (1969). Polym. Mech. (USA) 5, 824-826 (1972).
214. Kh. Akimbekov, V. S. Kuksenko, S. Nizamidinov, A. I. Slutsker, A. A. Yastrebinskii:
Fizika Tverdogo Tela 14/9, 2708-2713 (1972). Soviet Physics-Solid State 14/9,2339-2343
(1973).
215. R. V. R. Subramanian, T. V. Talele: Textile Research J. 42/4, 207-214 (1972).
216. H. M. Heuvel, K. C. J. B. Lind: J. Polymer Sci., A-2, 8,401-410 (1970).
217. G. Reinisch, W. Jaeger: Faserforschung und Textiltechnik 19/8, 363-365 (1968).
218. H. Kriissmann, G. Valk, G. Heidemann, S. Dugal: Angew. Chemie 81/6, 226-227 (1969).
219. H. Wilski: Physik-Grundlage der Technik (Physik 1974-Plenarvortrage), Weinheim 1974:
Physik-Verlag, p. 183-198.
220. T. M. Stoeckel, J. Blasius, B. Crist: J. Polymer ScL, Polymer Physics Ed. 16, 485 -500
(1978).
220: After completion of the manuscript this detailed study of the relation between
viscometry, ESR, and IR estimates of bond rupture has appeared. The authors report
that in PE and PP fibers and films very little chain scission can be detected through
viscometry; the number of broken bonds, N(vis), was 0.4 '10 16 to 5 '10 16 cm- 3 .
N(vis) is by a factor of 1000 smaller than the elsewhere reported IR end-group con-
249
centrations. Virtually no change of M was observed in cold drawing of a PA 6 mono-

filament whereas the viscosity-average molecular weight of PA 6, PA 66, and PETP
fibers decreased measurably during loading to fracture. Evaluating their viscosity and
ESR measurements the authors concluded that N(vis) and N(ESR) are of the same
order, if nonrandom scission prevails. In the case of random scission a number
N(vis) of broken bonds was derived from viscometry which was by a factor of 6 to
20 larger than the value N(ESR) obtained from the intensity of the ESR spectra.
Observing opposite influences of an annealing treatment of PA 6 fibers on micro-
crack formation and free radical intensity the authors question the close correlation
between microcrack formation and bond rupture.
250
Chapter 9
Molecular Chains in Heterogeneous Fracture
I. Fracture Mechanics. 252

A. Stress Concentration 252
B. Subcritical Crack Growth. 259
C. Critical Energy Release Rates. 271
II. Crazing . 272
A. Phenomenology 272
B. Craze Initiation 276
C. Molecular Interpretation of Craze Propagation and Breakdown . 286
D. Response to Environment 290
III. Molecular and Morphological Aspects of Crack Propagation 293
A. Fracture Surfaces . 293
B. Notched Tensile and Impact Fracture 306
C. Fatigue Cracks. 310
D. Mechano-Chemistry . 312
In Chapter 8 principally the contribution of spatially homogeneously distributed

molecular rearrangements to fracture initiation was studied. The term spatially
homogeneous referred to the absence of flaws, inclusions, cracks, or notches of a
size sufficient to act as a stress concentrator. Under those conditions during an initial
phase of external loading damage development or growth is homogeneously distri-
buted on a macroscopic scale. Heterogeneous fracture now is defined as the converse
of homogeneous fracture or briefly as fracture through crack propagation. In this
case cracks, notches, inclusions, or accumulated crack nuclei act as macroscopic
stress concentrations and essentially confine further damage development to the
close proximity of the then existing defect( s). The phenomenon of crazing has been
included in this chapter because of the well observable structural irregularities and
despite the fact that with increasing stress new crazes can be formed at arbitrary
nucleation sites.
The development of fracture through cracks will be treated in this chapter fol-
lowing the fracture mechanics approach. In view of the extensive, competent, and
recent literature on this subject prime attention will be given to the effect of chain
properties (length, structure, strength) on crack propagation and on fracture me-
chanics parameters.
251
9. Molecular Chains in Heterogeneous Fracture
I. Fracture Mechanics
A. Stress Concentration
In Section III B of Chapter 3 Griffith's early considerations [(2)] were outlined. From
the balance of elastically stored energy (U) and energy to produce new surface area
('YcA) Griffith derived his well-known fracture criterion for isotropic materials con-
taining an elliptical crack oflength 2a (Eq. 3.13). In the last 50 years this fracture
mechanics approach has been systematically developed to account for partially an-
elastic and/or plastic behavior of solids for various crack and sample geometries, and
even for material heterogeneities: In all cases it has been - and still is - the aim of
the fracture mechanics analysis to derive generally valid quantitative criteria for
crack stability and crack propagation behavior. Criteria are being sought which are
as much independent of the state of external and internal stresses and of crack and
sample geometry as possible and which depend principally on material functions such
as material resistance R (Ri~ausbreitungswiderstand), critical stress intensity factor
Kc (kritischer Spannungsintensitatsfaktor), and elastic modulus E.
A detailed account of the development and of the present state of the theory of
fracture mechanics is given in the series Mechanics of Fracture [2]. From the extensive
general literature on this subject only a few works may be cited which deal with the
deformation and fracture of engineering materials [3] or specifically polymers [4-6]
and standards [7, 8]. For the determination of the material functions R and Kc three
experimental methods of controlled crack propagation are predominantly used
(Fig. 9.1):
Mode I, crack opening or tensile mode (einfache RiMffnung)
Mode II, sliding or in-plane shear mode
Mode III, tearing or antiplane shear mode.
F or details of these methods, their evaluation, and possible influences of sample
geometry on the resulting "material" functions the reader is referred to the general
literature cited [2-8]. In this Section the most frequently used crack opening mode,
namely mode (I), will be considered (Figs. 9.1 and 9.2).
For a uniaxially stressed isotropic, elastic, thin plate containing an infinitely
sharp edge crack a state of plane stress arises which has a singularity at the crack tip
[3-8], for r ~a the components of stress are expressed as
Mode I Mode II Mode III

Fig. 9.1. Experimentally used modes of controlled crack propagation: I crack opening or tensile
mode, II sliding or in-plane shear mode, III tearing or anti-plane shear mode.
252
0'0 Fig. 9.2. Tensile specimen with single edge crack.
o = -Kr- cos -VJ (1 - . -VJ·sm

Slll -3 VJ ) (9.1)
x .,)2 rrr 2 2 2
Kr - cos -VJ
o = --
Y..../2iiY 2
(1 + . VJ· 32 VJ)
Slll - Slll -
2
(9.2)
Kr . VJ VJ 3
= - - - Slll - cos - cos - VJ. (9.3)
T
xy ..../2iiY 2 2 2
The coordinates rand ¢ are indicated in Figure 9.1. The quantity Kr determines
the intensity of the stress field without affecting its shape. It is accordingly termed
stress intensity factor. The stress intensity factor takes into account both the external
(uniaxial) stress 00 and the crack length a:
Kr =oo..;;ra . f(alb) (9.4)
where f(alb) is a correction factor of the order of unity which derives from the finite
specimen dimensions and crack configuration [3-8]. The stress intensity factor can
be considered a much more general measure of the criticality of the state of a stressed,
cracked plate than 00 or a individually.
In the case of a thick plate the lateral contraction of the material at the crack tip
is hindered; a state of plane strain prevails and in addition to Ox and Oy a normal
stress 0 z exists:
(9.5)
In plane strain and for r <{ a the components of displacement in the x -direction
(u) and the y-direction (v) are obtained [6] as:
u = 2 Kr(1 +
E
v\1 2rr 2
v
r cos ~ (1 _ 2 + sin 2 ~)
2
(9.6)
v= 2Kr(1+V)V r sin~(2-2v-Cos2 ~). (9.7)

E 2rr 2 2
253
For ¢ = 7r Eq. (9.7) yields the crack opening displacement
2v= - 8K1vRi
E
- ( l - V )2 .
27r
(9.8)
Whereas the stresses show a singularity for r -+ 0 the displacements go to zero.

Through the introduction of a crack into a (thick) plate under constant uniaxial stress
00 the elastically stored energy in the vicinity of the crack tip is increased by the
finite amount W:
(9.9)
Concurrently work 2 W is expended by the external system to maintain the uni-

axial stress 00' The total elastic energy Ue decreases, therefore, by W. This apparent
paradox was pointed out by Eshelby [9]. If the crack is extended by b.a then the
stress field is shifted by b.a in the x-direction and its energy content increased by
oW/oa; an energy increment of the same magnitude is set free by the stress release
accompanying the failure of the highly stressed crack tip material. It is the latter
energy increment which is termed energy release rate G1 :
G1 = - aUe = 7ra 05 (1 _ v2 ) f(alb). (9.10)

Baa E(T, t)
For a thin plate (state of plane stress) the terms (1 + v2 ) and (1 - v2 ) in Eqs. (9.6)
to (9.10) have to be replaced by unity. In view of the discussion later it is important
to determine how the energy G1 is distributed around the crack tip. Following Irwin
[10] the elastic strain energy within a cylinder of radius rl around the crack tip is
(per unit of plate thickness):
W(rl) = 7ra205
2E
(5--
4
34 2) -arl.
-v-2v (9.11)
Whereas the energy content W(rd increases proportionally to rl the local energy
density w(r) decreases as 1/r; for large distances from the crack tip it attains Wo =
OS/2 E, the strain energy density of an isotropic, elastic solid. From Figure 9.3 it is
readily seen that increasing a by b.a = r 1 sets free an amount of stored energy G1 b.a
which is a factor of (8 - 8 v)/(5 - 8 v) larger than the elastic energy stored within a
cylinder of radius rl around the crack tip. This ratio is independent of the absolute
value of r 1 . In other words any crack propagation step in an ideal elastic material -
even the smallest - releases elastic energy in a material volume several times larger
than that immediately surrounding the crack tip (volume 7rb.a 2 B). Experimentally
it is seen that a triangular zone bounded by a plane connecting the crack tip with the
lateral face is unloaded. Increasing a requires in any event the breakage of primary
and/or secondary bonds and thus energy, say Bb.a'Yc. In order to obtain a strict
correspondence between Bb.a'Yc and Bb.aG1 one would require that the total of the
254
,,
,
"''''.... ...... .....
w(r)rrlla 2
- .... --
Fig. 9.3. Distribution of elastically stored energy in front of a plane crack

(mode I). GI Aa energy
release (per unit specimen thickness) through propagation of crack by
Aa, W(r) energy stored
within cylinder with radius r about crack tip, w(r) energy density.
elastic energy stored in the contributing volume region be used for the separati
on of
bonds traversing BLla. For polymeric solids the smallest conceivable step would
be
the scission of a single bond, i.e. a step width of Lla = 0.4 nm. The activation
volume
for bond scission is of the order of (j = 0.008 nm 3 . Such a volume element constitu
tes
only a fraction 0.125/1T of 1TLla 3 and contains about 0.015 of G Lla 2 • It must
I be
noted, therefore, that on a molecular scale - even in the absence of plastic deforma
-
tion - fracture propagation requires the release of energy from a volume at least
60
times larger than the activation volume for breakage of primary bonds across
the
newly formed surface area BLla.
Propagation of a crack occurs if the energy release rate GI is sufficient to account
for the material resistance R, i.e. for all modes of energy consumption related
to the
propagating crack. The condition for crack growth is written, therefore, as
GI(a, t) ~ R(a, t).

(9.12)
At low crack propagation rates kinetic energy terms in R can be neglected.

Then the material resistance will include the specific surface energy 2 r (to overcom
e
the cohesion of atoms or molecules across the newly formed surface area), and
the
energy terms Vre of elastic retraction of stressed molecules, VpI of plastic deforma
-
tion, and Vch of chemical reactions including bond scission. The release of internal
stresses (Ui ) or chemical reactions with the environment (U ) may subtract
ch from R:
R=2r + oVre + OVPI+ OVCh+ ... _[OU i + OUCh].

Baa Baa Boa Boa
(9.13)
Boa
255
In general - and especially with polymeric materials - R will be a function of

crack length, specimen geometry, and extent and rate of plastic deformation at the
crack tip. A schematic representation of an R(a) of a ductile, strain hardening
material is given in Figure 9.4. A crack ao grows if Gr (ao) > R(ao); if G1 (ao) = R(a o)
it continues to grow if the slope of the line GI(a), i.e. 3Gj3a, is larger than the change
of material resistance with crack length, 3Rj3a. For an arbitrary energy release rate
GI with R(ao) < GI(ao) < Gc(ao) crack stability will again be achieved at a new
value at >ao where G(a) intersects R(a) (cf. Fig. 9.4). If, however, GI(a) is equal to
Fig. 9.4. Schematic representa-

tion of crack stability in ductile,
strain-hardening material:
0 0 a, 1. G < R(ao), 2. G(ao) = R(ao),
crack length 3. G(at) < Gc , 4. G = Gc.
or larger than Gc(a) - which forms the tangent to R(a) - then the rate of energy release
remains always larger than the material resistance and the crack will grow continually.
It follows from Figure 9.4 that Gc generally is the larger the larger R(a ~ ao). Since
R depends most strongly on the plastic energy term 3Vp1 !B3a it also depends on the
size of the volume element which deforms plastically during crack propagation. The
material resistance R is the smaller, therefore, the more constraint to plastic deforma-
tion exists, i.e. it is smaller in plane strain than in plane stress. Under conditions of
plane strain Gc approaches a limiting value of G Ic which - through Eqs. (9.4) and
(9.10) - defines a critical value of KI . This critical value, KIc ' is termed fracture
toughness. For a brittle fracture the numerical value of KIc can generally be deter-
mined from the measurement of crack length and load at the onset of unstable crack
propagation [3-8].
Plastic deformation is especially pronounced in polymeric materials. The electron
micrographs presented in Chapter 8 provide ample evidence of this fact. It will be
necessary, therefore, to study whether and how the fracture mechanics approach
derived for an elastic material can be applied to elastic-plastic solids as well. The
effect of plastic deformation on the stress distribution at a crack tip is well docu-
mented [3-7]. An elastic-plastic material under quasi-elastic conditions, for instance,
begins to deform plastically wherever the state of stress meets the yield or flow
criterion. Plastic deformation begins in the region of largest stresses, i.e. in the prox-
imity of the crack tip; it limits the components of stress to the yield stress of. In
256
1. Fracture Mechanics
order to maintain the mechanical equilibrium stresses must be raised (up to the yield
stress) in more distant regions. Thus the plastic deformation has the effect of in-
creasing the effective crack length [3-7]. Two common approaches to calculate the
effective crack enlargement through plastic deformation are based respectively on the
von Mises yield criterion [6] and on the consideration of aF as an additional com-
pressive stress [7].
From the von Mises yield criterion together with Eqs. (9.1) to (9.3) one derives
the shape of the plastic zone (r < rF) for a state of plane stress as [6]:
Ki
rF S = - - cos 2 -cf> ( . 2 cf»
1 + 3 sm - (9.14)
, p 2 rra 2 2 2
F
and for plane strain as
Ki
rF v = - - cos 2 -cf> [ ( 1 + 3 sm
. 2 -cf» - 4 v ( 1 - v)] . (9.15)
, 2 rra} 2 2
The plastic zone extends in the plane of the crack by
(rF,Sp)",=o = 2 Ki
rra 2 for plane stress; (9.16)
F
and by
(rF,V)",=O = - -Ki2 - (1 - 2 v)
2
for plane strain. (9.17)
2 rra F
In a moderately thick plate one finds a state of plane strain in the center and of
plane stress at the lateral faces. This leads to the well-known dog-bone shape of the
plastic zone.
The Dugdate model [7] recognizes the effect of the plastic deformation in front
of the crack tip in a different way. It considers a virtual crack which includes the
plastic zone - of size rp - and it treats the yield stress as a superimposed stress acting
normally on the boundary between the elastic and the "plastic" zone. The superim-
posed stress tends to bend the surface of the virtual crack. A consideration of me-
chanical equilibrium gives [6, 7]
2
r p -- -rraa
-2 0 ' (9.18)
8 aF
KI = a0 v''-1r7"(a-+-r-
p )'- (9.19)
and a crack opening displacement
2va=--
Kr (9.20)
EUF
257
Equations (9.14) to (9.18) do not constitute a fracture criterion since they do

not specify whether and when material separation in the plastically deformed zone
occurs. In order to use fracture mechanics to predict the stability of an elastic-plastic
crack, limits of the plastic deformation have to exist and must be known explicitly
or implicitly. Understandably the determination, interpretation, and application of
critical stress intensity factors for viscoelastic solids is particularly complicated due
to the pronounced time and temperature dependence of the mechanical properties
of these materials. The fact that a material shows a ductile behavior does not preclude
the use of fracture mechanics but it reduces - or even eliminates - the independence
of the fracture mechanics functions (G, K, R) from the geometrical parameters [6].
The occurrence of plastic deformation in a material has a threefold effect:
it limits the elastic energy density to u}/2 E
it leads to an increase in R (roughly in proportion to the volume of the plastically
deformed zone)
it may render uncritical any KI (Le. any Uo and a) unless the extent of plastic de-
formation is limited.
Following earlier concepts applied to metals the quantitative assessment of the
contribution of plastic deformation to material resistance R of polymers has made
considerable progress. Recognizing the difference between an Rl (brittle) observed
under plane strain and an R2 (plastic) resulting from the plastic deformation under
plane stress one may rewrite Eq. (9.12) as
(9.21)
Formally one may equate R2 with a GC2 ' the "plastic component" of the strain
energy release rate, and Rl with G CI which should correspond to G1c '
It has been indicated that Gc is the smaller the more pronounced the condition
of plane strain, i.e. the larger the specimen thickness B. At the other extreme of com-
paratively small thicknesses also a dependency of Gc on B is noticed (Fig. 9.5). The
B [mm]
Fig. 9.5. Gc as a function of specimen thickness for polycarbonate (from (211): (a) tapered slit,
crack length taken to be equal to visible crack, (b) tapered slit, plastic zone considered, (c) tapered
V groove, crack length taken to be equal to visible crack, (d) tapered V groove, plastic zone
considered; + constant thickness slit.
258
almost linear increase of Gc with B indicates that (in this case of a single groove double
cantilever beam) the volume of the plastically deformed zone (shear lips) increases in
proportion to B2. The decrease of Gc due to the beginning mixed mode fracture at
larger specimen thicknesses just becomes apparent [21].
At present several different experimental methods of working in the crack open-
ing mode are employed [7-15]. Flat samples with a single edge notch (SEN) are
preferentially used under constant load, under slowly increasing strain or under
cyclic tensile loading. These conditions permit a subcritical crack growth leading
eventually to instability. To determine the energy release rate Eq. (9.10) can be used
if the appropriate long-term relaxation modulus E(T, t) is employed. The same is true
for plates with a central notch (CN), compact tensile (CT) specimens, Charpy notched
samples in slow bending, and wedge open loaded (WOL) plates at small loading rates.
In shock loading the short-time moduli are relevant to describe the energy release
rate of pre-cracked samples; crack instability is reached primarily because of the in-
creasing stresses and at unchanged crack length. The contribution of thermally acti-
vated crack growth to an increase of KI is, therefore, negligible (the situation is dif-
ferent, though, with uncracked samples and flaws of molecular dimensions as has
been shown in Chapter 8).
The WOL and the double-cantilever beam (DCB) specimens are suitable to deter-
mine the arrest of an initially unstable crack; the double-torsion (DT) specimen per-
mits the investigation of crack propagation under constant KI [7-15].
Values of various types of energy release rates G and of fracture toughness values
K of polymeric materials have been gathered from the literature [12-75] and are
compiled in Table 9.1. Although the experimental method and the principal para-
meters and observations have also been indicated in Table 9.1, a discussion of some
molecular aspects will be attempted in the following.
B. Subcritical Crack Growth
The Griffith concept of an energy balance (G = R) furnishes a criterion of instan-

taneous stability (G < R) or instability (G > R, crack propagation). It does not
explicitely contain time as a variable. It has been pointed out above, however, that
G depends on t through E(t) and R depends on t predominantly through the plastic
energy dissipation term. The larger G, the rate of release of provided elastic energy,
the larger d, the rate of crack propagation. The opposite is not always true; thus
material resistances determined in bending, tension, or tearing may decrease with
the rate of deformation [17, 20-25, 35-37, 57].
In many polymeric materials three zones of crack propagation behavior can be
distinguished:
no crack growth at G < Gi
(stable crack)
slow crack growth at Gi < G < G1c
(subcritical crack growth)
rapid crack propagation at G ~ Glc .
259
~ Table 9.1. Energy Release Rates G and Fracture Toughness Values K of Polymeric Materials ~
0
~
Polymer Method and Specimen Glc * Klc* Parameters of Test, Observations Ref. (S'
(standard abbrev.) 11m2 MNm- 3 / 2 '"

=
Pr
n
Thermoplastic Materials it
5'
PE '"
5'
(LOPE) impact 5000 G CI (64) ::t:
(35000) ~
G C2 I1l
(medium density) impact 1300 (64) 0
G CI 0<1
CD
11900 G C2 ::s
(medium density) impact 14200 plastic zone correc~ion applied (69) 8
(p = 0.95 g cm- 3 , SEN 1.7-4 linearly increasing with "time to crack [57) '"
=
pigmented) initiation" (inverse of load rate) ~
10
(p = 0.96 g cm- 3 ) mode III tearing 2· 10 4 -20. 104 shows maxima and minima which are related (24) Sl
Gm ....CD
to the a- and /3-relaxation maxima =
(PE hard, black) SEN,CT 3600-20000 3.0-6.5 Kc (70)
PP
(lCI, compression SEN,SN 6.5-3.6 KC2 decreases with T, (-180 < T < 20°C), change in (68)
moulded) slope at -60°C related to /3-relaxation peak
2.8 KCI essentially constant
(BASF Novolen CT 5.2-4.5 v=0.08mms- 1 -40°C<T<20°C 1183)
1120 Lx) 3.3-3.75 v=8.6 mms- l
PBU 1
(non-modified) SEN 2.9-1.2 Kc gradually decreasing with "time to crack [57)
initiation"
(pigmented) SEN 1.2-3.2 distinct maximum of Kc at "times to crack [57)
initiation" around 0.1 to lOs
Table 9.1. (Continued)
Polymer Method and Specimen Glc * Klc* Parameters of Test, Observations Ref.
(standard abbrev.) J/m 2 MN m- 3j2
PS
SEN 1700 ± 600 (34)
DCB 300-2500 increasing with decreasing a [17,35-37)
(Shell Chemical) DeB 200-7.5 rapidly decreasing with increasing degree (45)
of orientation
SEN 0.6-2.3 10- 7 < a < 10- 2 mls [511
tapered cleavage 3.7-7 10- 9 < a < 10- 2 mls [511
(HIPS) impact 350 GCI (64)
(900) GC2
impact 1000 GCI •
15000 GC2
PVC
(BP, England) Charpy notched 2.3-6 -200 < T < 50 DC (63)
three point bending 2.2-5.6 T, rate of bending
(Cobex) DCB 3600 undrawn sheet (46)
DCB 850-50 decrease of Gc with increasing orientation
(1.2· 10- 3 < bon < 3.5 . 10- 3 )
impact 1230 GCI (64)
1440 GC2 (64)
lCI Darvic DEN ~
1.94-2.57 Kc 1711
(modified) impact 10000 Gc (64) ...>tj
'$;."
PMMA Ei
(l>
(lCI) SEN, DeB, impact, 1-3.5 -197<T<2l DC (14) ;:;:

(l>
(")
three point bending 2· 10- 8 < a < 0.08 mls
distinct peak of Klc related to side chain b) ~
;:l
N
0\ ~.
relaxation
N
0\ Table 9.1. (Continued) ~
N
s:
0
(standard abbrev.) J/m 2 MN m- 3 / 2 "~
()
SEN 200 (33) [

::;
SEN 1.7 [52) '"
S·
(Plexiglas, Rahm) SEN 1.2-1.8 0.1 . 10 6 < Mv < 8· 10 6 [15,66) ::c:
Klc increasing with molecular weight .....
'"
(Plexiglas, Rahm) SEN 1000-15000 [50,67) ....
'"
Gd increasing with a and M 0
0<>
200 < a < 700 mls ::;
'"
0.1 . 10 6 < Mw < 8.10 6 '0"
cleavage 120-600 various [17,21,35- fii
37,45,48) '"I1
....
III
237 GA (48) g.
~
(Perspex) (DCB) 100-35 decreasing with increasing orientation (45) ....
4 . 10- 4 < .1.n < 14 . 10- 4 '"
(plexiglas G) 0.75-100 log G1c increases linearly with log My (65)
(2 . 10 4 < My < 10 . 10 4
120 ± 20 Glc essentially independent of molecular weight
(My> 10 . 10 4 )
(Perspex) CT 1.3-0.75 decreasing with increasing T (74)
(Perspex) DT 1.6-1.7 (13)
(Simplex) 1.8-2.1 (13)
1.9-2.9 various a, T (12)
impact 1000 GCl (64)
1300 GC2
(perspex) 1.7-2 notch tip radius [58)
ABS
impact 12000-16250 Gc increasing with decreasing specimen width [53)
impact 1000-50000 steep increase of Gc between -20 and + 10°C (64)
impact 23000 (69)
(Lustran 244) impact 47000-49000 Gc [641
Polymer Method and Specimen Gic * KIc* Parameters of Test, Observations Ref.
(standard abbrev.) 1/m2 MN m- 3/ 2
PETP
(medium Mv) eharpy notched 4300 Gc is calculated from impact resistance of [53J
(high Mv) 5200 differently notched specimens [53J
(fiber grade) 7200 Gc values for amorphous and crystallized fiber [53J
grade PETP are the same
SEN, DEN 6.1-5.3-3.8 KC2
5.1-4.6-3.7 KCl crystalline, 20°C, -50°C, -100°C [49J
PC
(Makrolon) three point bending 5000 3.8-2.5 decreasing with increasing rate of deflection [22J
(10- 6 < v < 10 m/s)
(Makrolon) three point bending 10-2.5 decreasing with increasing temperature [54J
(-200 < T < -100 °el, slight maximum at 70°C
(Makrolonl impact 4-2 decreasing with increasing T (T < 50°C) [54J
2-15 increasing with increasing T (T > 50 °C)
(Makrolon) impact 3500 G CI [64J
5000 G C2 [641
4000 Gc calculated from impact resistance [531
(Lexan 101) 1500-2700 3.8-2.3 distinct maximum of G c at -40°C; [62J
decrease of Kic between -60 and +20 °e; ~
correlation of G c with tan Ii and fracture surface '!j
....
P>
morphology (")
[21,20]
...
cleavage 8000-170 Gle decreases rapidly with a E1
(1)
(0.25 . 10- 4 < a < 10- 2 m/s) ~
(1)
(DeB) 170-12 GIc decreases gradually with a [211 (")
::r
(10- 2 < a < 300 m/s)
t-J 8.
'"
0\ 1;l
W
tv Table 9.1. (Continued) ~
0\
s::
0
""" Polymer Method and Specimen GIc * Parameters of Test, Observations Ref. ~
KIc* (")
(standard abbrev.) 11m2 MN m- 3/ 2 s::

rrn
(Makrolon) DT 2.6-3.8 various temperatures and crack speeds 119]
(Makrolon) SEN 2.2 KCl (corresponds to Kc of crack initiation) [19] r-5·
8-5 KC2 decreases with increasing T
(-140 < T < -30 DC)
::r:
("i)
.....
(1)
...0
Polyamide 66 0<>
(1)
::;
(Maranyl) three point bending 8800 4.4-2.3 gradually decreasing with higher rates of [22] (1)
0
deflection ~
(Maranyl, SEN 4-7 Kc increasing with T (-160 < T < +40 DC) [68] ...'"I"l
~
0.2% H2 O) ~
SN 4 constant in same temperature region, s::
...
Kc
3-4 KCl little variation with T '"
8-10 KC2 maxima corresponding to 'Y- and ,,-relaxation peaks
are apparent at -140 and -80 DC
(Nylon 66, dry) impact 250 G Cl [64]
4150 G C2
(Nylon) impact 5000-5300 Gc [64]
Thermosetting Resins
PUR
(unsaturated impact 1300 Gc [53]
vinyl-urethane)
UP
(non-modified) SEN 440 1.2 Gc [56]
cleavage 42 ± 6 [59]
EP
(non-modified) DT,CT 0.55-1.7 -48 < T < +77 DC, various rates of deformation [75]
(non-modified) SEN 700 1.54 Gc [56]
cleavage 80 ± 5 [59]
(standard abbrev.) 11m2 MN m- 3 / 2
(non·modified) DCB-tapered 50-200 adhesive layer between aluminium substrates [73]

DT 210 ± 48 [411
(rubber-modified) cleavage 310 ± 10 [59]
(silica-filled) 2.2 ± 0.1
Phenolics
(phenol- SEN,CN 0.42-50 G[c calculated from rupture stresses according [55]
formaldey hde) . to Eq. :, G[c decreases with T
(20 < T < 205°C)
Composite Materials
Bakelite SR + Charpy 20000 - 36000 Gc values depend on fiber surface treatment [47]
40 Vol. % aligned high (HN0 3 , silane, brominated, silicone oil treat-
modulus carbon fiber ments resulting in inter laminar shear strengths r
(Morganite l) between 12 and 30 MN/m2
8800 morganite treatment with r = 55 MN/m 2
Epoxy + aligned SEN 5000-50000 13.5-19
RAE carbon fibers vf = 0.4) [56]
Gc and Kc for various crack geometries
10000-15000 20-28 vf = 0.6 and rates of deformation
UP + randomly SEN 8000-19000 vf = 0.15
oriented E glass fibers t""
Epoxy + boron fibers SEN 35000 47 vf = 0.66 ~
Plate Glass SEN 8.3 0.8 [6,49,60] ~
E;
(1)
~
(1)
("J
g'
tv ::;
0\ ~.
Vl
IV
0\ Table 9 .1. (Continued) ~
0\
~
Polymer Method and Specimen Glc* Klc* Parameters of Test, Observations Ref.
(standard abbrev.) J/m 2 MN m-3/ 2 '~"
(")
Elastomers (-
PBU S·
(mixed cis-1,4,trans- mode III tearing 18-1000 GIll increasing with effective rate of tear aTa [281 ::I:
~
1,4, and 1,2) (10- 22 < aTa < 10- 14 m/s)
40-78 threshold fracture energy To increasing with
Mc (2400 < Mc < 50000); Xs2TO independent
of test temperature, degree of swelling, type s::
1
of swelling liquid '"
(cis-1,4) mode III tearing 30-1000 GIll for various load rates, degrees of crosslinking, [281 ifa
and swelling
58 ± 8 threshold fracture energy for Mc = 3200 ~
81 ± 8 threshold fracture energy for Mc = 14000
(polysar) mode III tearing 1000-3000 Gill (for trousers and angled test pieces alike) [261
increases with rate of crack growth
(10- 7 < a < 10-4 m/s)
(butadiene- mode III tearing 100-100000 Gill increasing with rate of tear [251
styrene and (10- 7 < a < 0.1 m/s), decreasing with
butadiene- temperature after passing through a maximum
acrylonitrile
copolymers)
(polyurethane- pure shear 2.45 intrinsic fracture energy derived from visco- [291
solithane 113) elastic analysis of crack propagation
* As indicated in the fifth column, not always critical values Glc or Klc are listed.
This behavior is especially well demonstrated by PMMA. In Figure 9.6 the data
of Doll [30] are represented.
The slow subcritical crack growth in a polymer is due to thermomechanically
activated processes of the same nature as described in the last chapter for the homo-
geneous fracture, i.e. chain stretching, plastic deformation, void opening, chain dis-
entanglement, and chain scission. The rate with which these processes occur is, how-
ever, in the strain field surrounding a crack tip larger than elsewhere in the material.
[m/s]
~
I
I
i~
I
I
I
PMMA
<D (My,= 100000)
<V (Mw =2000000)
10-6 i I I, /' ..
~ ~ ~
Fig. 9.6. Crack speed as a function of energy release rate GI lafter (301).
The growth rate d in the subcritical region seems to be determined by K1 . (In

subcritical crack growth any imposed KI corresponds per definitionem to the Kc for
crack growth at the observed speed.) Several relations between KI and have been a
established. One of these has been found to describe the sub critical crack growth of
a number of polymers [3, 12, 13]:
d=AK~ (9.22)
with A and b being empirical constants, depending on material, environment, tem-

perature, etc. The constant b has been related to l/tan 0 of the {3 relaxation [12].
The subcritical crack growth data obtained by several investigators [12-15] for
PMMA have been compiled in Figure 9.7. The double-logarithmic plot reveals an
extended linear section (growth rates d from 10-8 to 10- 2 m/s for stress intensity
factors from 0.8 to 1.4 MN m -3/2).
For this linear portion the exponent bin Eq. (9.22) can be determined to be
25 ± 2 [13].
From Eqs. (9.22) and (9.4) one can determine the time .l t necessary for the
growth of a crack a 1 to length a2. Neglecting here and in the following f(alb) one
obtains for 00 = constant:
a2 da KI2 2 K dK 2(K11
2- b _ K 12
2- b)
.It=f -.-=f 1 1 (9.23)
al a Kll o~ 7r AK~ (b - 2).l7ro~
267
----- -v--~
PMMA
v, 1
5
m " ......
v
- ;.
v
10-2 - ,- ",
v /,0 + Beaumont
o Marshall et 01·
v Radon et 01.
S Doll et.al.
K 1c (water)
Fig. 9.7. Slow crack propagation in PMMA

D.S 1.2 1.6 2.0 M N 1m 312 at room temperature (after [111, including
stress Intensity factor data from ref. 112-15 n.
If one choses as upper limit Kic and as lower limit KIm' i.e. that value of K which
corresponds to the largest flaw or defect present in the sample, then one obtains as
~ t the lifetime tb of the sample under constant stress. The extremely strong expo-
nential dependence of ~ t on KIm makes tb very sensitive to flaw size. Increasing the
size of the largest flaw, am, by only 10% decreases the lifetime tb by a factor of 3.
The above considerations on the dependence of tb on the size of inherent flaws
have been extended by various authors [13,16-17] to a method of predicting the
creep life of glassy polymers. It is assumed that the size am of the largest intrinsic
flaw can be determined from the breaking stress ab in short time loading according
to Eq. (9.4):
(9.24)
Broutman and McGarry [17] thus obtained intrinsic flaw sizes of 50 Jlm for
PMMA and of 620 Jlm for PS. Berry [13] reports values of between 62 and 95 Jlm
for PMMA. These values seem to be very high and no direct optical observations of
such flaws have been reported. They are also two to three orders of magnitude larger
than the diameters of microcracks found in PMMA and listed in Table 8.3. The micro-
cracks are assumed to have dimensions comparable to those of a globular substructure
which may be responsible for their appearance under stress. The "intrinsic flaws"
must have a different significance, therefore, and may only develop at a stress level
close to abo
The higher the crack propagation rate in a sample gets (d > 10-: 3 m/s) the more
the relation between d and KI - and even the measured value of Kic - depend on
the parameters of the experiment. As shown in Fig. 9.7 Kc increases monotonically
up to a Kic of 1.6 MN/m- 3 / 2 if ao is kept constant [12, 13]. In other experiments
(Charpy impact tests, crack initiation and arrest in DCB specimens) K-values at higher
crack speeds (2 to 300 m/s) are obtained which lie below the Kic [14, 18].
268
The different points on the d - KI curve in Figure 9.7 represent quite different
energy release rates G1. According to Eqs. (9.4) and (9.10) G1 increases with the
square ofK1. For the material employed one may use an average elastic modulus E
of 4.5 GN/m 2 and a v of 0.36 thus obtaining G1 values of 70 J m- 2 at the lower
range ofKc (0.6 MN m- 3j2 ) and 500 J m- 2 for a K1c of 1.6 MN m- 3j2 • These G
values are equal to the corresponding material resistance R under the given experi-
mental conditions. Their significance and interpretation will be discussed in the fol-
lowing Section.
The slow crack growth behavior of poly carbonate, also a glassy polymer, is even
more complex than that ofPMMA. At low temperatures (T < -40°C) Kc values be-
tween 2.6 and 3.4 MN m- 3 / 2 have been obtained [19] from DT specimen which
were independent of crack speed at small crack speeds (d < 1O- 3 m/s) but dependent
on sample thickness and temperature. At higher crack speeds (d ~ 10- 1 m/s) Kc
values slowly increased. Double-side grooved DCB specimen (DG-DCB) however
showed a decrease of Kc with increasing Ii in the region of d < 10- 1 m/s [20]. This
behavior was confirmed by Kambour et al. [21], who observed a Gc of 8.2 kJ/m 2 at
d = 2.5 . 10- 5 m/s and of 12 J/m 2 at d = 300 m/s. The microphotographical investiga-
tion of the slowly fractured sheets (B = 12.7 mm) revealed the - energy-consuming-
formation of shear lips 0.4 mm wide (mixed mode crack propagation). At high speeds
no shear lips were found. By copolymerisation of PC with silicone blocks the authors
[21] were able to increase the fracture toughness in the whole temperature range
T > -110°C. With this material mixed mode crack propagation occurred fairly
independently of crack velocity.
The increase of crosshead speed (over 7 decades) in static three point bending of
PC resulted [22] only in a slight decrease of the calculated Kc from 3.8 to 3.0 MN
m- 3 / 2 • From the three point bending test a K1c is derived [14] from:
(9.25)
if P is taken as the largest load measured, ac/aa as the proper change of specimen
compliance with growth of a (perpendicular) crack, and BN as the thickness of the
load carrying section. The quantity ac/aa increases with increasing crack length.
At the beginning of crack growth the values of P and ac/aa (and the corresponding
Ku) will be smaller than the Klc at catastrophic failure. Arad et al. [22] point out
that Kn and K1c are the closer together the less material ductility and/or slow crack
growth are favored by the experimental conditions.
In relation to studies of the fracture surface energy as a function of loading rate
the early and wide application of the tear test (mode III) must be mentioned
(e.g. [5, 23-28]). In this fracture mode the material in the crack tip zone is sub-
jected to a complex and highly plastic deformation. Without going into details it
may be mentioned that rate effects on the extent of plastic deformation (and thus
on fracture surface or tearing energies) have been found [23-29] and correlated
with (3- and 'Y-relaxation maxima [5, 23-26]. The tearing energies of thermoplastics
and rubbers are generally quite large: e.g. up to 1 kJ/m 2 for PS, 20-200 kJ/m2 for
PE, and 0.1-500 kJ/m 2 for various butadiene copolymers [24-26]. For elastomers
269
Thomas [27] and Ahagon and Gent [28] report that after adjustment for the chang-
ing effective fracture area a common threshold fracture energy To of 40 to 80 J/m 2
could be determined under different experimental conditions. This energy was shown
to be independent of temperature and degree of swelling for various types of swelling
liquid. The threshold energy decreased slightly with increasing degree of crosslinking
(of the polybutadiene samples). In aggressive environment (oxygen, ozone) To is sub-
stantially reduced.
An experimental and theoretical analysis of the steady propagation of a crack in
a viscoelastic continuum is due to Knauss [29]. Employing a polyurethane elastomer
(solithane 113) as material Knauss studied crack propagation under pure shear. From
his extensive analysis, Knauss derives a solution to the viscoelastic-boundary-value
problem represented by a crack moving in an isotropic, homogeneous, incompressible
solid. He finds that the stress singularity at the crack tip vanishes. Under these condi-
tions the stress-intensity factor only describes the far-field loading conditions. Knauss
observes that the rate-dependent fracture energy is essentially the product of an
"intrinsic fracture energy", presumably of molecular origin, and of a non-dimensional
function which relates to the rheology of the material surrounding the crack tip. For
the polyurethane elastomer this intrinsic fracture energy amounted to 2.45 J/m 2 • As
criteria for crack stability are concerned, he found that the criterion of ultimate,
constant strain (crack-opening displacement) is as applicable as a generalized energy
criterion.
As stated in the introduction to this section the subcritical crack growth in a
polymer is due to the thermomechanical activation of different molecular deforma-
tion processes such as chain slip and orientation or void opening. The energy dissipated
depends on the frequency, nature, kinetics, and interaction of these processes. But
there are many and notable attempts to treat the sub critical crack propagation as
one thermally activated multi-step process characterized by one enthalpy or energy
of activation and one activation volume. Several of these kinetic theories of fracture
have been treated in Chapters 3 and 8.
The concept of thermal activation has been extended to the fracture mechanical
analysis of crack propagation. Thus Pollet and Burns [179] have discussed earlier
approaches and the significance of various variables. Commonly the energy release
rate G1 is considered as the mechanical intensive parameter. In their treatment, the
crack front is thus represented by a line on which a force per unit length, G1 , tends
to move it in the forward direction; its motion is, however, restrained by the presence
of "thermal" obstacles, i.e., obstacles, or barriers, which can be overcome by thermal
activation. Whether the physical origins of those energy barriers are individual ob-
stacles placed on a unique surface along the fracture path or whether they are located
somewhere in the surrounding region, does not affect the generality of the treatment
if G1 can be considered as an intensive variable.
With the assumption of thermally activated barriers, the average crack velocity
can be written according to a general Arrhenius-type equation as:
II = Vo exp [-LlG(G 1, T)/kT] (9.26)
where the quantity Vo can be thought of as the maximum attainable crack velocity
and LlG as the free enthalpy of activation to overcome one "obstacle".
270
As observed by various investigators [12-15,180] an approximately linear rela-

tion holds between the logarithms of crack velocity and crack extension force:
(9.27)
Pollet and Burns find the data for PMMA taken by Atkins et al. [180] at various
temperatures to comply with Eq. (9.27). They determine from C(T) an area of activa-
tion A* = C(T)/G1 of the order of 0.7 . 10- 4 to 2' 10- 4 nm 2 . They interpret this
small value as indicating that the thermal obstacles are not lined up on a single surface
along the fracture path but distributed in space in a region of material near the ap-
parent fracture path. In view of the consideration in Section IA it should be said that
G1 is not an intensive variable. Only a small fraction a of G1 contributes to a reduc-
tion of the energy of activation. If slow crack propagation were due solely to the
repeated shear displacement ofPMMA segments of 0.2 nm length over a distance of
0.8 nm an activation area of 0.16 nm 2 would have been expected. Since 2 . 10- 4 nm 2
have been formally calculated one would have to conclude that a amounts to
2· 10- 4 /0.16 = 1/800.
C. Critical Energy Release Rates
It has been discussed in the previous Section that an increase in stress intensity factor
or in G1, the driving force for crack extension, promotes sub critical crack growth
(Figs. 9.6 and 9.7). Since the material resistance R was found to increase with d a new
equilibrium between G1 and R could be established following any change in G1. If
G1 of KI are continually increased, however, a point of crack growth instability is
reached. The instability may be characterized by the statement that at this point the
material resistance R(d) as given by Eq. (9.13) is not sufficiently rate sensitive to
compensate the increase in G1. Consequently the crack accelerates to a level of
velocities where inertia forces and the finite speed Ve of elastic waves have to be con-
sidered [67,181-182]. So far the contribution of the kinetic energy of the receding
fracture surfaces to R has been neglected. At the point of incipient crack instability
in PMMA with crack velocities of about 0.1 m/s the kinetic energy term amounts to
6 J /m 3 • At these velocities this term is, therefore, an insignificant fraction of the
average strain energy density of 100 to 500 kJ/m 3 • At velocities of between 100 and
1000 m S-I, however, an increasingly larger fraction of the stored elastic energy
G1(static) is necessary to account for the kinetic energy of the crack front - reaching
100% at a crack speed equal to the Rayleigh surface wave velocity VR' The dynamic
strain energy release rate Gd is, therefore, only a fraction F of G1(static) :
Gd = 1Ta~a(1-v2)F{'
E \a, Ve
)
(9.28)
Broberg [181] derived the function F(d, ve) for an infinite, isotropic, elastic plate.
His expression may be approximated by (1 - d/VR)'
271
Similar considerations apply to the slowdown and eventual arrest of a rapidly

propagating crack [18, 182]. An understanding of this phenomenon is of consider-
able importance in the use of high pressure pipes where the propagation of a crack
over long distances must be prevented under any circumstances.
The transition from subcritical to rapid crack propagation is especially pronounced
in PMMA but more or less clearly observed in all polymers capable of brittle fracture.
The corresponding critical quantities K1c and G1c characterize the fracture behavior
of a material. In Table 9.1 an (incomplete) listing of measured £ritical values
(G1c and K1c ) and of energy release rates referring to particular experimental condi-
tions (G c ' Gd , GIll) are given. The main experimental parameters and their ranges of
variation have been indicated.
Of the various terms of Eq. (9.13) contributing to the material resistance the
plastic deformation, aVpdBaa, is generally by far the largest. It has been attempted
in Table 9.2 to identify also surface tension 'Y and the (hypothetical) energy density
of chain breakage UIqN L' The smallest term is that of the specific surface energy 'Y
(free surface energy, surface tension). The 'Y values of polymers lie in the range of
0.020 to 0.046 Jim 2 [31]. The energy of elastic retraction is the product of strain
energy density and width of the retracting layers. If the layers consist of fully
oriented chains stressed to break and thus having the highest possible energy density
(of the order of 1000 MJ/m 3 ) the energy of retraction of chain ends of length
L = 5 run amounts to 8 J/m 2 .
The chemical surface energy, aVch/Baa = Uo/qN L also contributes comparatively
modestly to R, even if one chain breakage event per molecular cross-section q is con-
sidered: 0.4 J/m 2 in PMMA, 2.9 J/m 2 in HDPE. The terms Ui and Uch cannot be
evaluated in all generality. The physico-chemical data used for the above and sub-
sequent calculations are collected in Table 9.2 for seven semicrystalline or glassy
polymers.
II. Crazing
A. Phenomenology
The phenomenon of crazing has been observed in many glassy polymers, and also in
some crystalline polymers, when subjected to tensile stress. Polymer crazes (Fig. 9.8a)
in their appearance are similar to the very fine cracks known for a long time to occur
on the surfaces of inorganic materials such as ceramics. However, there is a difference
between the crazes and the cracks in that crazes have a continuity of material across
the craze plane (Figs. 9.8b, 9.9-9.11) whereas cracks do not possess any continuity.
Consequently crazed zones are capable of bearing loads as opposed to cracked ones.
In the last thirty years a growing number of publications has been devoted to the
phenomenon of crazing. Two comprehensive reviews [76-77] on this subject have
appeared in 1973. The bibliography on crazing compiled in this monograph [78-178]
is mostly to be understood as an extension of the earlier reviews.
272
Table 9.2. Molecular and Mechanical Data of Selected Polymers (after [11,31], Average Values for Non-modified, Unoriented Polymers at
Room Temperature)
Quantity Symbol Unit HDPE PP PA 66 PS PMMA PVC PC
Density Q g/cm 3 0.96 0.91 1.14 1.06 1.19 1.39 1.20

Young's modulus E MN/m2 1200 1300 1300 3100 3300 2600 2350
"Chemical surface energy" U/qNL J/m2 2.9 1.3 1.6 0.5 0.4 0.5 0.6
I'exp J/m2 0.031 0.029 0.046 0.040 0.039 0.039 0.042
Surface tension
I'theor = {j 2~ J/m 2 0.029 0.052 0.084 0.071 0.102 0.058
Monomer molecular weight Mm g/mol 28 42 226 104 100 62.5 254
Activation energy for
Uo kJ/mol 335 272 188 230 175 (84) 117
bond scission
Monomer length 1m nm 2.53 2.17 17.3 2.21 2.11 2.55 10.75
Monomer cross-section q = Mm/lmqNL nm 2 19.0 35.1 18.9 73.2 65.7 29.1 32.8
Solubility parameter {,2 MJ/m 3 268 354 (774) 332 502 427
Undisturbed statistical chain
ho 10 3 M- 1/2 A. (mol/g) 1/2 950 765 890 670 640 720 880
end-to-end distance
Chain rigidity s = ho/hof 1.63 1.61 1.63 2.3 2.14 1.83 1.1
......
......
Q
N N
'"
-...l ~.
W
Fig. 9.8. Crazing of polytrifluoro chloroethylene under constant load (from [130 I): (a) appear-
ance of crazed specimen surface, (b) electron micrograph of a craze showing fibrillar craze
material which connects rather straight boundaries.
The subject of crazing has been extensively researched for various reasons:
in brittle fracture of a crazeable material, cracks generally grow and propagate
through crazed material,
the formation of crazes under stress, and/or in presence of an active environment,
effect (and generally impairs) mechanical and optical properties, surface quality,
and permeability,
crazes trace and make visible the statistical and geometrical distribution of the
variables of their initiation (states of strain or stress, molecular properties, struc-
tural defects).
Crazes occur most commonly in amorphous, glassy polymers such as PS, SAN,
PMMA, PVC, PSU, PPO, and PC but also in semicrystalline ones (PE, PP, PETP, and
POM). Crazes have also been observed in epoxy [131] and phenoxy resins [l09].
With reference to the cited literature (e.g. [76-78, 85, 106]) it can be said that
crazes are sharply bounded regions fIlled with "craze matter". The craze matter
consists of oriented strands interspersed with voids and has an average density of 40
to 60% of the matrix density (Figs. 9.9 and 9.10). Kambour [76, 85] has described
the structure of crazes to be similar to an opencelled foam in which the average dia-
meter of the holes and polymer elements is about 20 nm. There is much evidence
that the polymer elements are oriented primarily along the direction of the principal
tensile stress. Beahan et al. [106] noted that a fairly uniform fibril structure exists
over the whole length of a tapered craze (Fig. 9.9). From the uniformity of this
structure, these authors conclude that the increase in the width of the craze with
sample extension is mostly due to fibrillation of further matrix material rather than
to an increase in fibril extension. More extensive studies [150] have shown, however,
that there is a change in the micromorphology of the craze along its length, with
coarser, thicker fibrils being found in the tapered section near the craze tip. The
morphology of crazes in thin cast films as well as in bulk specimens is rather similar
274
II. Crazing
Fig. 9.9. Newly formed craze in thin slice cut

from uncrazed bulk polystyrene; craze grow-
ing from left to right in a direction perpendic-
ular to that of the uniaxial tensile stress
(Courtesy D. Hull (1061).
Fig. 9.10. Electron micrograph of the central

section of a craze grown as that in Figure 9.9
(Courtesy D. Hull (106)) .
Fig. 9.11 . Schematic diagram of the variation

in polystyrene craze structure with increasing
craze width; the angle at the tip of the craze
is exaggerated and the scale in region d is
larger than that of a to c. (Courtesy D. Hull
d c a (1151).
b
[115]. In the bulk, fibrils of 25 to 50 run in diameter have been observed (Fig. 9.11,
b and c) . At low strain rates, the crazes grow in thickness by drawing further matrix
material but also to some extent by an increase in the orientation and elongation of
the fibrils leading eventually to the breakage of the fibrils. The first stages have been
shown schematically in Figure 9.11. The principal observations with regard to craze
morphology are indicated in Table 9.3.
The mechanical response of fibrils in the direction of the applied stress has been
determined for polycarbonate [83] and polystyrene [120] . Figure 9.12 shows a stress-
strain curve of craze material obtained by Kambour and Kopp [83] for Pc. It may
be noted that the craze material can sustain tensile stresses only slightly less than the
yield stress aF of the bulk material. The strains, however, are enormously greater in
the craze - between 40 and 140% as compared to the bulk yield strain of about 2%.
275
OY,eld ( PC )
50
MN/rri
Polycarbonate 0
40
bulk
30
..
..
~
20
10 20 30 40 'I, 50
Fig. 9.12. Stress-strain curve of craze material
craze strain in polycarbonate (after [831).
It follows from this behavior that the energy of deformation of a thoroughly crazed
sample can attain very large values (cf. 8 II A).
B. Craze Initiation
In spite of the fact that a considerable amount of research has been conducted on
various aspects of crazing, there has been no exact agreement on the mechanism of
craze initiation. So far, there is no theoretical model available which can predict in
all generality whether a certain polymer will craze or not under a given set of condi-
tions. And, if it does craze, what effect the temperature and rate of deformation will
have on craze formation and propagation. This is certainly due to the fact that the
initiation of a craze depends simultaneously on three groups of variables characterizing
respectively the macroscopic state of strain and stress, the nature of defects of hetero-
geneities in the matrix, and the molecular behavior of the polymer in the given thermal
and chemical environment. In focussing on particular variables five conceptually dif-
ferent approaches to describe the craze initiation emerge. These are based on stress-
bias, critical strain, fracture mechanics, molecular orientation, and molecular mobility
respectively. The principle observations and the craze initiation criteria proposed on
the base of the above concepts are listed in Table 9.4.
These criteria describe different states of local excitation and deformation of
chain segments. The stress bias criterion [86, 139] refers implicit ely to two mecha-
nisms: cavitation in a dilatational stress field and stabilization of cavities through a
deviatoric stress component. These mechanisms have been more explicitely con-
sidered in the mathematical model of cavity expansion in .a rigid plastic by Haward
et al. [137] and in the molecular models by Argon [152] and Kausch [11].
276
Table 9.3. Morphology of Crazes in Glassy Polymers
Observations Conclusions Material and

References
Optical and mechanical study of the effect of various Crazing is a mechanical separation of polymer chains or Hsiao and Sauer
variables, such as magnitude and duration of stress on groups of chains under tensile stress. Crazing does not (78)
initiation and development of crazing. occur in materials oriented along the draw direction. PS
X-ray scattering and electron microscopy suggests that Crazes resemble on open-celled foam, the holes and polymer Kambour
void content of the craze is dispersed in the form of elements of these average about 20 nm in diameter. (85)
inter-connected spheroidal holes having a common Craze-tensile behavior is an extension of the craze formation PC
dimension of 10- 20 nm. Stress-strain curves of crazes process. Decrease in elastic modulus and yield stress with
have been determin'ed showing a yield stress followed increasing strain are rationalized in terms of strain-induced
by a recoverable flow to roughly 40 to 50% strain at decrease in density and resultant increase in stress concen-
41 to 55 MN/m 2 . Returned to zero-stress, the craze tration factor on the microscopic polymer elements of the
exhibits creep recovery at a decelerating rate. craze. Polymer surface tension and large internal specific
surface areas of the craze are suggested to b'e important
factors in large creep recovery rates of craze.
Microstructure of crazes in thin films by transmission Fibrils of 25-50 nm are a common feature of the micro- Beahan, Bevis & Hull
electron microscope, structure of crazes in cast and bulky films. (115)
In thin PS films e (craze initiation) = 1% Major fibrils diameter = 20-30 nm. PS
Minor fibrils connecting major fibrils have a diameter
> 10 nm and they tend to orient normally to major fibrils.
At large total deformation at low E, there is a gradual transi-
tion from coarse to fine microstructure of crazes.
Optical and micromechanical study of PS craze behavior: The large number of parallel crazes formed permitted Hoare and Hull ...
- average craze thickness = 0.19 /-Lm the determination of the stress-strain curve of a craze; (120)
N spacing between the crazes = 38 /-Lm it is similar to that found by Kambour and Kopp (83) PS if
N
~.
:j - crazes start forming at 0.5 OF and e = 1.0% for solvent-induced crazes in PC (Fig. 9.12)
~ Table 9.4. Criteria of Craze Initiation ~
00
~
0
Observation Conclusion Materials ~
(")
~
References
rrn
Stress-based criteria [
~
Analysis of craze initiation stresses in tensile samples containing Stress bias criterion: I al - azl;;. A(T) + B(T)/I l Sternstein et al. S·
circular holes, in thin-walled biaxial tube samples, and for rods A, B material constants depending on thermal history [86.139] ::r:
('I)
in combined tension-torsion tests and environment, PMMA ....
('I)
..,
I 1 = al + a2 + a3 > 0; 0
O<l
('I)
hydrostatic pressure suppresses crazing ::s
('I)
0
Analysis of location and direction of crazes formed Crazes are initiated at the intersection of lines where shear Camwell et al.
[141]
5i
- in sheets compressed by a die or wedge bands are formed. Since crazes can be initiated before or ..,'rj
- during three-point bending of a round-notched bar after shear band formation it is assumed that large hydrostatic PS po
(")
tensions rather than strains cause the craze initiation. Critical Ishikawa et aL ..,2"
('I)
hydrostatic tensions: [172]
87 MN m- 2 for slowly cooled PC PC
89 MN m- 2 for quenched PC
Strain-based criteria
Analysis of location and direction of crazes formed in Criteria of principal strain and of strain energy agree best with Wang, Matsuo, Kwei
tension at the interface between steel balls in PS and test data, maximum principle stress, maximum principle shear, [105]
rubber balls in PS and stress bias criterion correspond less well, maximum dilation PS
the least
Craze formation in long-term tensile testing Time-dependent critical principal strain; Menges, Pohrt,
limiting values (t - 00), in %: Schmidt et al.
[104,110,113,1211
PS 0.3 [110] POM 1.5-2 [104,1211
SAN 0.7 [110] PP 2 [121]
PMMA 0.7-0.9 [104,110] PC 2 [110]
PVC 0.8-0.9 [104]
Craze formation in biaxial stress tests "second quadrant" €c = A'(T) + B'(T)/I l Oxborough et al.
[142]
PS
Observation Conclusion Materials

References
Fracture mechanics criteria

Study of craze behavior in sharply notched PMMA Craze growth follows one of two distinct patterns. One leads Marshall et al.
immersed in methanol to eventual craze arrest, the other to a constant speed of [102)
Craze initiation and growth characteristics are controlled craze propagation and final failure. PMMA
by the initial stress intensity factor Ko not by applied stress: Kinetics of craze growth in methanol explained by liquid
Ko < Km no crazing flow through the porous crazed material (to which a void
Ko < Kn craze initiation and arrest spacing of 0.25 /lm, a void size of 72 nm and a craze yield
Km = 0.06 MN m- 3 / 2 stress of 9 MN m - 2 are assigned).
Kn = 0.2-0.3 MN m- 3 / 2 (depending on sample thickness)
Study of crack arrest in tensile loading of cast PMMA sheets In a plot of a;(square of the threshold stresses for crack Andrews et al.
in various organic liquids propagation) versus reciprocal crack length a a distinct lower [124)
bound was found which defined a critical energy release rate PMMA
Gj; whereas Gj initially decreases with temperature it becomes
independent of temperature at T > T c where T c roughly
corresponds to the T g of PMMA in the particular environment.
Photo elastic study of craze initiation and propagation in Stress concentration at crack tip is sensibly reduced with craze Narisawa et al.
kerosene and near a Griffith crack. Initial value of Ko is initiation and not transferred to craze tip. [127)
the governing factor for craze initiation and propagation There is a critical stress intensity factor Km below which no PMMA
x = At n crazing occurs: PC
where x = craze length Km = 0.13 MN m- 3 / 2 PMMA
A = constant depends on Ko Km = 0.29 MN m- 3 / 2 PC
n = 0.53 for PMMA The difference in sorption kinetics is held responsible for the
= 0.24 for PC higher critical stress intensity factor and the lower exponent
obtained from craze growth histories n in PC, relative to PMMA. ...
N
t?
N
--J
\0 JJ'
~ Table 9.4. (Continued) :0
o
~
~
Observation Conclusion Materials (")
t::
References
fr
n
Structural and morphological criteria
[
~
Study of craze initiation stress ai, of location of crazes within Craze initiation occurs in regions where chains were oriented Haward et al. S·
bulk samples and of change of birefringence R across sample transversely to direction of stress (birefringence positive) [89] ::c:
~
thickness x PS
aj = 46 MN m- 2 for dR/dx = 0
aj = 38 MN m- 2 for dR/dx = 2 . 10- 3
~::s
g
Study of birefringence and of tensile and compressive The crazing stress increases with amount of orientation when Hull et al. ~
behavior of hot-drawn sheets at various angles to draw the tensile axis is parallel to the draw direction, and decreases [1531 ~
direction with amount of orientation when the axis is normal to the PS "'g.
draw direction. I=i
(1)
At high draw ratios the crazing stress parallel to the draw

direction is higher than the shear yielding stress and the
material undergoes a yielding process similar to compression
yielding.
Molecular orientation has only a small effect on the orientation
of the craze plane.
The effect of molecular orientation on the crazing behavior
of polystyrene is primarily associated with the nucleation
stage of the crazing process.
Statistical nature of initiation of stress crazing is explored; The time required for the formation of crazes is an inherent, Narisawa et al.
a plot of histograms of the time to craze initiation follows but statistical, characteristic of the material itself involving [118]
approximately an exponential distribution. a stress depending rate process reflecting to some extent the PMMA
initial microscopic flaw distribution.
Observation Conclusion Materials

References
Molecular criteria
Relation between yield stress uF and craze initiation stress Both yield stress and .:razing stress vary with temperature, Haward et al.
Uj is investigated: but crazing stress varies less than the yield stress. This 11081
uF(T) E = constant indicates that crazing has to provide the nucleation surface PS
Uj (T) E = constant energy for voids.
where Uj decreases less than uF at high temperature. Similar response of crazing and yielding data to temperature
indicates that in both cases similar molecular conformational
A[ISOaUF ] [aUj] changes and backbone motions are involved.
- - - and - - -
a(log E) a (log e) Materials having highest crazing tendency (PMMA, PS, SAN)
are represented by the same function f(T) also have high yield stress and thus a large energy associated
with local plastic instability (which is necessary to account
for the nucleation and growth of the tiny voids).
Analysis of craze initiation in tubular specimens under Crazes initiate at surface or interface stress concentrations Argon et al.
combined tension and torsion considering the mechanism where localized plastic flow produces micro cavities by intense 1152,1661
of craze growth as one in which craze tufts are produced inhomogeneous plastic shear at a molecular scale. PS
by repeated break-up of concave air-polymer interfaces The rate of cavity formation depends primarily on the activation
at the craze tip (meniscus instability). free energy ~G* (cf. Eq. 8-30) and on the local concentrated
For a surface energy 'Y =0.05 11m2 and a tensile plastic deviatoric stress while the subsequent plastic expansion into
resistance of u y = 10 2 MN m- 2 , an extension ratio craze nuclei is in response to global negative pressure.
An = 2 of craze matter gives a craze tuft diameter =
0.05/.Lm.
-
~
N
N Jg0
00
The strain criteria [104, 110, 113, 121] specify that no crazing will occur if the
largest principal strains - even at extended times - remain below a limiting value,
which was found to be some 2% or less. At short times craze initiation strains may,
in some polymers, be of the order of 3 to 5% [104,113]. In the absence of intensive
strain concentration strains of this order are insufficient to cause chain scission but
they tend to decrease the intermolecular attraction (cf. 8 II B). This is recognized
by Oxborough and Bowden [142] who also use the tensile hydrostatic stress (Id in
their general critical-strain criterion. In biaxial stress fields the stress bias criterion
and their criterion differ only little; as Oxborough and Bowden point out Sternstein's
data and their own can be explained by either one of the two criteria.
If one would apply either of the above criteria the craze initiation in a sharply
notched plate loaded in tension, one would have to expect instantaneous crazing
since both I a 1 - all and E show a singularity at an infmitely sharp crack tip
(cf. Eqs. 9.1 to 9.3). Such an expectation would be contrary to the experimental
findings. Marshall et al. [102] and Narisawa et al. [127] have established that it is the
initial stress intensity factor Ko which controls craze initiation at the crack tip. In
the case of PMMA and PC immersed in methanol or kerosene critical values Km exist
below which no craze initiation and growth occurs. This behavior can be understood
in view of the discrete sizes of the chain segments and of any voids to be formed, in
view of the fact that the density of stored elastic energy is limited (Fig. 9.3), and in
view of plastic deformations eliminating the stress singularity. Marshall et al. [102]
conclude from their data that crazing occurs when the material at the crack tip
reaches a critical strain or crack opening displacement.
The effect of sample structure and morphology on craze initiation has been
recognized since the earliest investigations [78]. The structure of the sample surface
is for various reasons especially important:
- defects and/or contaminations are preferentially found at the surface enhancing
craze initiation there [173, 176];
injection molded or machined samples mostly contain a surface structure which
differs from that of the bulk material [89, 175, 176];
molecular chain segments in surfaces have a higher degree of mobility [112];
environmental attack (diffusion, plastification, degradation) begins at the surface
and penetrates from there.
The investigations of the effect of chain orientation on craze initiation show that
a transverse orientation of the chains with respect to the direction of principal stress
enhances craze initiation [89, 153]. Because there are fewer chain segments pointing
in the direction of principal stress critical local strains are obtained at smaller stresses
(cL 3 IV E). On the other hand the craze initiation stress increases with the degree of
alignment of the chains in the stress direction (increasing degree of orientation, small
angles 8 between draw direction and principal stress). In the case of good alignment
craze initiation stresses will be higher than the stress for shear yielding so that no
crazing is observed. For PS at 20°C yielding in tension occurs in samples drawn to
/. . = 2.6 or more and at a 8(/...) smaller than 20 to 30° [153]. Particularly noteworthy
is the observation of Hull and Hoarse [153] that the molecular orientation has only a
small effect on the orientation of the craze plane.
282
II. Crazing
This indicates that the transfer of stresses along the axes of the oriented chain
segments does not determine the direction of craze extension.
As an effect of sample structure and morphology the enhancement of craze
initiation through surface flaws [118,152,173,176], impurities [155,161], and
heterophase inclusions (reviewed in [114, 168, 190, 191] has also to be mentioned.
The molecular criteria generally take into consideration molecular weight, chain
entanglement and the local mobility of chain segments at the given temperature and
chemical environment [11, 15,50, 79, 146, 165-167,173]. The effect of the chemical
environment will be discussed in Section D. The follOwing investigations generally
have been conducted in standard atmosphere. An interpretation of craze initiation
in terms of local strain softening has been given by Rusch and Beck [95]. They pro-
pose that there exists a critical strain for crazing which is dependent on the level of
frozen free volume initially distributed in the bulk of the material. Through the im-
posed dilatant strain sufficient free volume is added to bring the polymer to a state
similar to the one reached by it at its T g' Gent [96], too, suggests that craze initiation
results from lowering of the Tg in local high stress fields at surface flaws. However,
as shown by various authors [89, 129, 158] surface crazes occur in PS, PMMA, and
PC at temperatures of -75°C and below. This raises a doubt whether the polymer
reaches a state characteristic of its Tg at such low temperatures or not.
Lipatov and Fabulyak [112] point to the importance oflow temperature relaxa-
tion processes due to the motions of side chains. These relaxations shift towards
lower temperatures in samples with high surface to volume ratios. This behavior has
been interpreted in terms of lower segmental packing on the surface and, therefore,
easier molecular motions. It is claimed that this facilitates crazing. Shifts of molecu-
lar relaxations towards lower temperature (in PC) have also been observed by Sikka
[163] who suggested that this shift may have its origin in the formation of microvoids.
Studies of molecular weight dependence of crazing and of breaking stresses have
been carried out by Fellers and Kee [146]. These authors suggest that craze initiation
breaking stress
----~I----~.~-,~-------.
crazing stress
2Me
O~~ __- L__ ~__L-~ __-L__~__~~~~~~~ 4

3 4 5 6 7 8 9 10 11 12 13 Mn 14 15·10
molecular weight
Fig. 9.13. Breaking stress and crazing stress as a function of molecular weight Mn of polystyrene
at 25°C (after (1461). Me entanglement molecular weight.
283
stresses are independent of molecular weight when Mn > 2 Me whereas craze devel-
opment and breakdown are clearly molecular weight dependent. This is evident from
the plot of stresses versus molecular weight, shown in Figure 9.13. Since the measured
elastic moduli (1.5 GN m- 2 ) did not depend on molecular weight the craze initiation
strains are also constant (2%). From these results it may be concluded that craze
initiation is an event which depends primarily on the interaction between chain
segments. Contrary to the apparent independence of stress and strain at craze initia-
tion to molecular weight the T g of the different PS fractions rises monotonically
from 88°C at Mn = 70000 to 105°C at Mn = 150000_ Another observation made
by these and other authors is that crazes in high molecular weight samples are very
fine, numerous, long, and straight as compared to the ones in lower-molecular weight
polymers, which are coarser in texture and somewhat shorter in length. This pheno-
menon seems to be related to the homogeneity of the stress field which is the better
the more numerous and the stronger the fibrils. Based on these studies and on the
extensive investigations of Doll et al. [15, 30, 50, 60, 61, 67] a model of craze initia-
tion and propagation is presented (Fig. 9.14). The effects of molecular weight
vm d
0
B,
I
-I I
I A
0 10 20 30 ~m
Fig. 9.14. Model of craze initiation and propagation in PMMA (after (111): A. elastically strained
region. B. region of void nucleation and fibril formation, C. craze growth due to fibril extension
and continued fibrillation at fairly constant stress Oy. D. transformation of craze into crack
through breakage of stressed fibrils.
(Mn > 2 Me) on crazing may be characterized using this model:

Mn has very little influence on the elastic straining behavior (region A) and on the
stress or strain at which crazes are initiated (B)
after craze initiation, fibrils are formed which are stronger and which can develop,
under increasing stress and time, to greater lengths the higher Mn ; consequently
region C is the more extended the higher Mn
the force transmitted by region C grows with Mn and so does the stress at break.
284
II. Crazing
In an attempt to reconcile the various partially insufficient craze initiation

criteria with the experimental data Argon [165-167] and Kausch [11] have proposed
models of the craze initiation process which take molecular structure, chain rigidity,
conformational changes, and intermolecular interaction into consideration. Argon
bases his criterion of the transition from A to B on the break-up of concave air-
polymer interfaces (cf. Table 9.4). Kausch has described a three-step craze-nucleation
mechanism:
a primary step of de cohesion of molecular coils with low degree of entanglement,
i.e. the formation of unstable micro-voids (B 1 in Fig. 9.15).
a secondary step of anelastic deformation of adjacent molecular coils containing
topological entanglements (B2 in Fig. 9.15),
a third step of direct lateral transmission of strain onto adjacent coils which leads
to a preferred formation of micro-voids in correlation to the first .
. ~
f?2
,,
/}';
,
-f--
/lnql)1I3
B1
\
, TI
\ '1~/6)"2
.
lOnm
Fig. 9.15. Model of void nucleation and fibril formation in PMMA (after [Ill): A. elastically
strained region, Bl' possible void nucleation at site of insufficient penetration of molecular coils
1 and 2; shaded areas indicate zones of strong intermolecular interaction and transfer of stresses
onto neighboring coils 3, B2' possible fibril consisting of interpenetrating coils 3; if their response
is of sufficient amplitude, void nucleation is furthered in adjacent sites (between coils 4 and 5).
The micro-voids are stabilized through the anelastic deformation of coils in the
secondary step. According to a suggestion of Fischer the attainment of a critical
macroscopic compliance permitting an anelastic deformation of the coils of sufficient
amplitude is the necessary prerequisite for the secondary step to take place [11].
With sustained or increased load the micro-voids coalesce and leave micro-fibrils
285
which connect the opposite craze surfaces. These micro-fibrils are formed, therefore,
of moderately elongated, physically entangled molecular coils. The diameter of the
micro-voids of 10 to 20 nm and that of the fibrils of 10 to 50 nm indicate that the
formation of a micro-void is an event involving only very few molecular coils.
C. Molecular Interpretation of Craze Propagation and Breakdown
The foregoing discussion on the phenomenology of craze initiation has also given a
molecular interpretation of the propagation of an individual craze. Taking into con-
sideration the formation of many crazes either simultaneously, or in sequence, a par-
ticular craze grows through:
displacement of the craze tip into uncrazed material by the formation of voids
ahead of the craze tip or by coalescence with a separate and smaller craze,
widening of the craze while extending the craze matter, Le. the molecular strands;
depending on the overall conditions strands may rupture or disentangle giving rise
to a crack.
There are three aspects of craze propagation which will be discussed at this point:
the kinetics of craze growth, the stress at the craze-matrix interface (region C in
Fig. 9.13), and the craze breakdown.
The rate of longitudinal craze growth has been studied by a number of authors.
The many problems still existing with regard to the transition of matrix material
into craze matter and with the rheological properties of the latter are found to be of
double weight in any quantitative description of craze propagation. For this reason
no detailed description of the various approaches will be given here - but the basic
concepts should be mentioned. The fracture mechanical investigations on PMMA
[15,50,102,127,133] and PC [127, 144] have led to empirical expressions for
d(a + rp)/dt, the rate of craze growth, in terms of stress intensity factors. Kambour
[76] and Marshall et al. [102, 133] stress the importance of environmental flow
through the porous craze material. Verheulpen-Heymans [155] formulates a model
for craze growth based on stress and strain analysis around a craze and on the rheo-
logical properties of the craze matter. In those cases where the craze length was
found proportional to the crack length [15,144,177] the empirical law for crack
growth (e.g. Eq. 9.22) also describes craze growth.
The stresses at the craze-matrix interface may be determined from the craze
shape. It was found by Weidmann and Doll [15] for PMMA and by Fraser and Ward
[177] for PC that the shape of these interfaces conformed very well to the predicted
shape according to the Dugdale-Muskelishvili plastic zone model [7]. In that model
it is assumed that a normal stress equal to the matrix yield stress aF is acting on the
interfaces. This stress aF is derived from Eqs. (9.18) and (9.20) as 1TEva/4rp. Using
the approximately molecular-weight independent ratio of rp/2va ~ 27 as given by
Weidmann and Doll [15] for PMMA and using an E of 4.5 GN m- 2 one arrives at an
aF of 65 MN m- 2 . This value corresponds to the strength of the isotropic PMMA
(Table 1-1). Fraser and Ward [177] obtained in the same wayan interfacial stress of
58 MN m- 2 (at 22°C) and 130 MN m- 2 (at -130°C) for PC.
286
II. Crazing
Another method, Knight's analysis of stress distribution along a craze, has been
used in a number of investigations (cf. [76,77]). Verheulpen-Heymans [157] quite
recently pointed out, however, that the - mostly unknown - rheological behavior
of craze material and of the craze tip zone has such a strong effect on the calculated
stress field that at the present time the results of this method cannot be interpreted
unambigously.
Using holographic interferometry Peterson et al. [148] determined the distribu-
tion of the incremental stresses caused by a strain increment applied to a craze in PC.
At low prestrains (1.3-1.7%) they observed that the craze material near the tip of
the craze actually supporte~stress increment .6.ay well in excess of the average
stress increment .6.a y (.6.a y / .6.a y equal to 1.2 to lAS). The peaks in the stress incre-
ments occurred at a distance of 0.2 to OA of the craze length behind the craze tip.
In their photoelastic studies of PMMA and PC crack tips Narisawa et al. [127] noted
high stress concentrations just before the initiation of a craze and very modest stresses
60 min later. No noticeable new stress concentration appeared at the craze tip.
In a growing craze molecular strands are formed (at B in Fig. 9.14) and extended
(zone C in Fig. 9.14). The growth of a craze is slowed down or even halted
if the craze hits upon an obstacle which prevents further fibrillation of matrix
material (e.g. a heterophase inclusion; a zone of higher orientation, larger flexibility
or lower stresses; the end of the specimen)
if the craze has grown to such an extent that the initiating crack or flaw is blunted
and rendered uncritical
if the craze interacts with other crazes or with the environment in such a way as
to change the conditions which have led to its initiation and/or growth (stress
relaxation of the sample because of craze-related creep deformation, penetration
into a lower-stress zone adjacent to another craze).
On the other hand craze growth will continue or be resumed if the molecular
strands fail (transition from craze to crack, zone D in Fig. 9.14). It becomes imme-
diately clear from Figure 9.14 and Eq. (9.18) that the energy dissipated in the open-
ing up of a craze is the higher the longer and wider the craze (zone C) and the larger
aF. By no means all questions are answered concerning the structure and rheology
of the molecular strands and of the interface craze-matrix. Nevertheless it can be said
that at a void content of 50% a molecular strand is subjected to a tensile stress of
about 2 aF, a stress which does not vary too much with development of the craze
from B to D. The mentioned maximum in incremental stress at a distance of 0.2 to
OA of the craze length in PC does not contradict the above statement. It indicates,
however, that the differential compliance of a freshly developed strand is smaller
than that of an extended fibril.
While being subjected to the tensile load the molecular strand obviously degrades
and eventually fails and this causes a transition from a craze to a crack. The break-
down occurs by one or more of the following mechanisms:
slippage and disentangelment of molecular coils leading to fibril rupture (frequently
of the central section of the fibril),
fibril rupture through breakage of entangled chains,
fibril-matrix separation.
287
The phenomenon of the loading and breakdown of the molecular strands has
been studied by various methods. Optical and electron micrographs of crazes are
shown in most of the cited references. Examples have been reproduced in Figures
9.8-9.10. Investigations of the craze shape by interference microscopy [15, 155, 177]
have also been discussed. At this point some results will be reported which have been
obtained by thermal measurements [31, 50, 184-186], from analysis of the influence
of molecular weight on crazing [11,15,65,79,146,178], through acoustic emission
[174,188], and by the ESR technique [189-190] respectively.
Doll [31, 50] determined the amount of heat Q dissipated in PMMA at the tip
of a rapidly propagating crack. To derive Q they measured the temperature rise
ilT 1(t) at a thermocouple placed on the specimen surface close to the prospective
fracture plane. Considering geometry, thermal conductivity, and crack speed athey
calculated the heat Q dissipated per unit are of crack surface. As expected Q increased
with ti and Mw (from 0.15 to about 5 kJ/m 2 ) and it was found proportional to the
dynamic energy release rate G d . Extrapolating Q(G d ) down to G1c (at Ii = 140 m S-1)
they observed that the dissipated heat Q (li = 140 m S-I) = Qo accounted for only
57% of the released mechanical energy G1c ' They ascribe this difference to the fact
that the highly stressed molecular strands conserve an amount of energy of the order
of 0.43 G1c , that this energy is released at the moment offibril rupture and that it is
dissipated over wide distances in the form of small shock waves which escape measure-
ment by the employed difference technique [30]. The acoustic emission analysis
(AEA) of a crazing PMMA sample by Roeder [188] permits similar conclusions. The
formation of crazes (transition A ~ C in Fig. 9.14) did not give rise to any discernible
acoustic signal. Whereas the transition from a craze to a crack (C ~ D) could be traced
by AEA even at an earlier stage than by optical microscopy. In PS, however, craze
formation is accompanied by sharp cracking sounds at strains of 0.6% or less [78].
The thermal measurements are not only interesting in view of the energy balance
of the crazing process but also because they permit the calculation of the local tem-
perature rise il T0 caused by the opening up and breakdown of the craze in PMMA. Doll
[30] assumed that Qo was initially confined to the zone of crazed material. Using a density
of 0.6 g cm- 3 , a specific heat of 1.46 J g-1 K- 1, a craze layer thickness 2 v of
1.65 pm, and a Qo of 335 J m- 2 he arrives at a ilTo = 230 K. This value is in accord
with theoretical estimates of Weichert and SchOnert [185] and with infrared measure-
ments of Fuller et al. [184] on PMMA. The latter determined within a range of ti of
200 to 640 m S-1 a constant ilT of 500 K. The Simultaneously observed increase
in Q(d) implied that the plastic deformation at the crack tip became more extensive
at higher crack speeds. Preliminary experiments on PS showed a ilT of 400 K and
smaller heat values [184]. These temperatures are certainly large although not com-
pletely unreasonable. They indicate that not only melting but thermal degrada-
tion must occur under these conditions. And they are in proportion with the much
higher temperature rises - of several thousand K - which were deduced by Weichert
[186] from the spectroscopical analysis of light emitted during fracture of glass.
It was noted at an early stage that sample molecular weight is an important
variable in craze propagation and breakdown. Rudd [79] studied the stress relaxation
ofPS fllms in contact with the crazing environment butanol. A plot of a(t)/a(O)
versus logarithm of time revealed that the time to reach the 40% stress level was three
288
II. Crazing
orders of magnitude larger for the higher molecular weight samples. The final decay
from 0.4 0(0) to zero stress occurred rather rapidly for all molecular weights [79].
The observation of Fellers and Kee [146] that the breaking stress ofPS only
gradually increases once Mn > 2 Me has already been mentioned in the previous
section (Fig. 9.13). Their results conform quite well to those of Doll and Weidmann
[15,50]. These authors determined the shape of a craze, the released heat Q and the
material resistance R for a series of PMMA samples having well defined molecular
weights Mw between 1.1 . lOS and 8 . 106 • Measuring the crack opening 2 v, the
craze width 2 ve ' and the craze length rp at a crack speed of 10-8 m S-1 they noted
that these parameters of the craze shape increased with Mw up to an Mw of about
2· lOs. At higher Mw hardly any changes of 2 v and rp and very small increases of
craze width were observed [15]. This means that initially (Mw < 1.6 . lOs) craze
widths increase with chain lengths. The craze widths are found to be by a factor of
5.2 larger than the extended chain length. This does not mean, however, that each
molecular strand consists of just a few, highly extended chains. One may assume
that before craze initiation the molecules have their randomly coiled conformation.
Chains ofMw = 1.1 . lOS for instance have an end-to-end distance of 21 nm. During
craze opening this distance will on the average increase by the fibril strain, i.e. to
about 30 nm. A total of 2360 of these elongated coils are contained in a fibril of
20 nm diameter and 1200 nm length. At molecular weights comparable to Me the
interpenetration and entanglement of these coils hardly allows the formation of
fibrils [11, 146,187]. At higher molecular weights (up to Mw = 2· lOS) the coils are
larger in size and the absolute number of entanglements per coil increases; the fibrils
become stronger and more resistant to failure by slippage or disentanglement. At
even higher molecular weights (Mw > 2' lOS) and under constant load the fibril
strength seems to be more determined by the strength than by the flow behavior
of the molecules. Under oscillating loads disentanglement may even occur at such
molecular weight levels. Recent experiments of Skibo, Hertzberg, and Manson [191 ]
strongly support the contention that the breakdown of molecular strands as a con-
sequence of their disentanglement triggers the stepwise growth of fatigue cracks.
This phenomenon will be discussed in more detail in Section 9 III C.
The critical energy release rates GIe show a behavior similar to that of the craze
shape. For small Mw the GIe are strongly dependent on Mw' Thus they increase from
1.4 J m- 2 at Mw = 2· 104 to 110 J m- 2 at Mw = 12 to 15 . 104 [65]. At higher
molecular weights - up to 8 . 106 - only a gradual increase of GIe to values of be-
tween 160 and 600 J m- 2 is observed [30,65]. This corresponds to what has been
said above on the effect of molecular weight on fibril length and strength.
The occurrence of chain rupture in craze breakdown has been inferred from the
above thermal, mechanical and acoustical evidence. Its direct observation by ESR
technique is difficult because of the limited number of crazes which can simulta-
neously be carried to breakdown. The number of chain breakages generally corre-
sponds to that of a single fracture plane. No quantitative data on free radicals formed
in crazing have been reported. DeVries et al. [189] have observed a very weak ESR
signal in a crazed and cut PMMA fiber. Nielsen et al. [190] did not find any evidence
offree radicals in thoroughly crazed - but unbroken - samples ofPS and ABS.
289
The above considerations mostly dealt with an individual craze, with the condi-
tions of its initiation, propagation and breakdown in a brittle polymer. Generally a
larger number of crazes is formed in a stressed specimen (cf. Fig. 9.8a). If those
crazes are sufficiently distant from each other then they will grow freely. Opfermann
[175] found that in PMMA a lateral distance between crazes of 80 J1ffi assured their
unimpeded longitudinal growth. The macroscopic mechanical properties (strain at
break, strength at short or long times, energy to break) depend to some extent on
the number of crazes per (surface) area, but they are still comparable to those of a
brittle solid with a strain at break of some 4 to 5% and a low fracture energy. In order
to sensibly increase the macroscopic creep compliance and the energy to break, crazes
must be initiated in large numbers throughout the sample volume and they must be
prevented from premature breakdown. Both objectives have been achieved with
heterophase copolymers or blends.
The obvious technological importance of the toughening of brittle polymers
through controlled crazing is reflected by the considerable amount of work done in
this area. In a recent monograph by Manson and Sperling [192] a comprehensive
elaboration of the physical, chemical, technological, and materials science aspects of
polymer mixtures, rubber-toughened plastics, block copolymers, grafted copolymers,
interpenetrating networks, and polymer blends has been given. The reader is referred
to this monograph for any detailed information on these systems. The conference
proceedings [193] from the Bad Nauheim meeting on multiphase polymer systems
also treat synthesis, compatibility, and mechanical properties of copolymers and
blends. In addition, references are given which focus on individual aspects. Argon
[152] and Retting [168] have reviewed the crazing of heterogeneous glassy polymers.
A quantitative method to describe creep and impact toughening of HiPS is presented
by Bucknall et al. [114]; the effect of crazing on the stress-dependence of creep in
PS and SAN is discussed by Moore et al. [94]. the surface morphology of stress
whitening in ABS by Morbitzer et al. [97], the effect of rubber on the environmental
stress-crack resistance ofPE by Spenadel [117], and the rapid healing of crazes in
ruptured rubber-toughened ABS/MMA sheets by Takahashi [143].
Valuable information on the role of crazing in breakdown of a polymer is ob- .
tained, of course, from fractographY. Relevant investigations [30, 49, 61, 66, 132,
150, 155, 169, 194-204] on this subject will be discussed in Section III A.
D. Response to Environment
In the previous sections on crazing primarily mechanical and molecular parameters

have been discussed while the chemical environment was not considered as a variable.
At this point an overview over the physico-chemical response of a crazable material
to an active chemical environment will be given. (The chemical response to an active
physical environment such as photo degradation or ozonolysis has been treated or
referenced in 8 III).
The physico-chemical actions of a (gaseous or liquid) chemical agent on a
polymer involve adsorption and absorption of the agent, the swelling and/or plasti-
cization of the matrix, the reduction of surface energies, and/or chemical reactions
290
II. Crazing
such as the hydrolytic depolymerization. These aspects of environmental stress

cracking (ESC) have been treated in numerous original articles and some reviews
(e.g. [80, 76-77, 171 D. From the extremely large body of experimental observations
only a few can be discussed in the following.
The physico-chemical actions of a liquid environment may influence the initia-
tion, propagation, or breakdown of a craze in a thermoplastic polymer. There seems
to be agreement that a liquid must be able to diffuse into a polymer in order to in-
fluence craze initiation. Narisawa [119] determined the critical stresses Uj for craze
initiation in thin films of PS and PC in contact with various alcohols and hydro-
crabons. He observed that crazes appear without significant delay and that ui
decreases with decreaSing chain length of the solvent (from 45 to 20 MN m- 2 for
PS, from 70 to 50 MN m- 2 for PC). From these results he concludes that a slight
swelling of a microscopic surface area is necessary and sufficient to effect craze
initiation. He derived a criterion for Uj in the form of Eq. (8.29) with activation
volumes of the order of 1.0 to .13 nm 3 , activation energies of 109 to 130 kJ/mol,
and rate constants of 1 to 10 . 10- 38 S-1 (PS) and 2 to 50 . 10- 45 S-1 (PC).
The solvent crazing in terms of polymer and liquid solubility parameters has been
studied in detail by Andrews et al. [124,126] and Kambour et al. [125, 128]. Kambour
employed a wide range of swelling liquids with solubility parameters Os between
5.34 and 19.2 cal 1/ 2 cm -3/2. He determined the equilibrium solubility Sy, expressed
as volume of liquid absorbed per unit volume of polymer, for PS, PPO, and PSU.
The swelling of PPO across the whole spectrum of organic liquids was found to be
inversely correlated by los -0 PPO 1and, consequently, the craze resistance was
correlated by 1os-oppo I. In PS and PSU craze resistance was less well correlated by
the solubility parameters. With all three polymers, however, the equilibrium solubility
Sy gave a good measure of the polymer-solvent interaction. Using Sy as independent
variable, unique plots of T g and of craze initiation strain €j were obtained for two
sets of polystyrene data. One set was obtained from samples preplasticized to various
degrees by o-dichlorobenzene; the other from "dry" samples in contact with the
swelling agent (€j) or swollen films (Tg). Kambour concludes from these results that
the presence or absence of a liquid/polymer interface is immaterial to the crazing
effectiveness of a given crazing agent.
The crazing agent thus acts through its presence within the polymer matrix. In
increasing the chain mobility (lowering Tg) it facilitates the primary and secondary
steps of craze initiation: nucleation and stabilization of a craze. This leads to the
lowering of Uj and €j in brittle polymers such as PS. Easier nucleation and stabilization
even cause the appearance of crazes in otherwise ductile materials such as PPO, PSU,
PVC, or PC.
Andrews et al. [124, 126] also studied the equilibrium swelling, of PMMA,
in various alcohols and related it to the observed changes in yield stress, UF, glass
transition temperature, T g, and material resistance R. They report the interesting
phenomenon that R above a certain critical temperature T c became independent of
temperature. Within experimental error T c corresponded to the T g of the PMMA
under the particular conditions of swelling. They propose a relation for R involving
the surface energy of the nucleated cavities and UF. They then interpret the constancy
of R(T > T c) with the vanishing of uF(T > T c). The absolute values of R(T > T c)
291
are rather small: about 0.1 J m- 2 (or PMMA in isobutanol, carbon tetrachloride,
n-propyl and isopropyl alcohols, 0.3 J m- 2 for ethanol, 0.45 J m- 2 for methanol
[124 ].
The swelling, crazing, and cracking behavior of PMMA, PVC, and PSU in contact
with some 70 different liquids was analyzed by Vincent and Raha [123]. They con-
sidered not only the solubility parameter Os but also the value of the hydrogen
bonding parameter HOD based on the displacement of the OD infra-red absorption
band in CH 3 0D in presence of the particular liquid and benzene. Better (but still
no unique), correlation between the mode of failure and Os and HOD was obtained.
An extensive investigation of the role of metal salts in stress cracking of various
polyamides was carried out by Dunn and Sansom [90-93]. Metal halides permitted
to distinguish two types of action: the formation of complexes between the metal
and the carbonyl oxygen and interference with the hydrogen bonding (found for the
chlorides of Zn, Co, Cu, Mn) or with solvent cracking, as with Li CI, Ca C1 2 , Mg C1 2 ,
or Li Br [90-91]. The action of metal thiocyanates on polyamide 6 was similar to
that of the corresponding metal halides [92]. Of various nitrates Cu(N0 3 h had the
strongest effect on the stress cracking ofPA6 fIlms [93].
In the investigations of Marshall and Williams et al. [52, 102, 133, 151, 164]
Narisawa et al. [127], Kitagawa et al. [144], and Krenz et al. [159-160] craze growth
and creep were related to fracture mechanical quantities. Graham et al. [164] model
crazes in PMMA as a line plastic zone subjected to a craze stress a er . In the presence
of active liquids aer is reduced from its air value of 100 MN m- 2 to e.g. 7 (methanol),
5 (ethanol and propanol) or 10 MN m- 2 (butanol). A clear linear correlation be-
tween Km and aer was found to exist. For the craze size at initiation (i.e. just before
the beginning of crack growth) the authors found a unique value of 11.5 f..LITl [164].
Similar fracture mechanical studies of solvent crazes in polystyrene were carried out
by holographic interferometry by Krenz et al. [159-160].
Studies of the micromechanics of deformation in semicrystalline polyethylene
in environmental stress cracking agents [107, 154, 170, 171, 204] have elucidated
the role of spherulites and of the mosaic block structure and its disintegration into
independent nonuniform fibrils. The morphology of crazes in amorphous PVC in
liquid and vapor environment was examined by Driesen [122] and Martin et al. [198].
It has not always been recognized that liquid gases such as N2 and Ar are also
acting as environmentally active agents. Parrish and Brown et al. [116, 129] have
been the nrst to report that the fracture stresses of PE and PTFE immersed in helium
at 78 K are respectively 21 and 33% larger than those obtained in liqUid N 2 . Later
studies of Brown et al. [134, 157-158], Peterlin et al. [135,137-138,145] and
Kastelic and Baer [140] concern N2 , Ar, O2 , CO 2 , and He and a variety of materials
(pP, PTFE, PETP, PC, PMMA). It is the common consensus that
- N2 and Ar exert an environmental effect on the tensile stress-strain curves of all
polymers at low temperatures
- the environmental effect is the larger the closer a gas is to its condensation point
and the greater its thermodynamic activity [145]
- N2 , Ar, O2 , and CO 2 usually cause craze yielding whereas in He or in vacuo the
polymers undergo brittle fracture without crazing
- the gases have to act at or close to a free surface
292
III. Molecular and Morphological Aspects of Crack Propagation
as crazing mechanisms gas absorption and matrix plasticization are proposed as

well as gas adsorption and surface energy reduction which facilitates cavity forma-
tion.
From the discussion in this section a number of craze suppression mechanisms
have become apparent which may be summarized here :
imposition of hydrostatic pressure in addition to tensile stresses [86,139,142]
introduction of compressive stresses into the surface [77, 174, 176]
orientation of molecules in the surface layer [89 , 153]
plasticization, preferentially of the surface [82, 162, 178]
coating of the surface by an oligomer of the sample material, e.g. coating of PS
by a 600 molecular weight polystyrene oligomer [178] polishing of the surface
[173, 178]
increase of sample molecular weight.
A. Fracture Surfaces
Fractography, the study of the morphology of fracture surfaces, is an obvious tool

to elucidate the origin and the mode of propagation of a crack. In this monograph
fracture surfaces of fibers (in 8 I G) and of other samples have already been analyzed
where appropriate. At this point some additional remarks are necessary concerning
particularly the effect of chain length and intermolecular attraction on fracture
surface morphology.
The discussion in 9 I C of the various terms contributing to the critical energy
release rate G1c and the data assembled in Tables 9.1 and 9.2 have made it clear that
a fracture surface is obviously not simply formed by the breakage of primary and/or
secondary bonds across a fracture plane of molecular dimensions. There is always a
plastic deformation of the crack tip zone and, consequently, of the ensuing fracture
surface. It is to be expected that the extent of plastic deformation is the smaller the
smaller the segmental mobility, i.e. the lower the temperature. At liquid nitrogen
temperature most high polymers resemble glasses and fracture in a brittle manner.
Fig. 9.16. Central section of a fracture surface

from a notched HOPE specimen broken in
bending at liquid nitrogen temperature
(from (1301) .
293
Viewed without magnification the fracture surfaces (e.g. Fig. 9.16) show a macro-
scopic roughness but they appear to be locally smooth although not shiny. This
indicates that the surfaces contain structural irregularities larger than the wavelength
of light. This is the case with e.g. PE, PP, PVC, PS, and also with PMMA which, how-
ever, has a very smooth surface.
The macroscopic features of the shown fracture surface (Fig. 9.16) can be ascribed
to the propagation of a cleavage crack at high speed normal to the direction of local
tensile stress. The local stress field is strongly affected by elastic waves generated at
earlier stages of crack development and by the initiation of secondary cracks. The
fracture surface was obtained by bending a notched polyethylene sample at liquid
nitrogen temperature [130]. The surface is locally smooth but otherwise full of steps
and ridges. The intersection of wavefronts and crack planes under different (e.g. right)
angles leads to the curious patterns as shown in the bottom right of Figure 9.17. The
characteristic properties of thermoplastics, molecular anisotropy and strong rate
dependency of deformation become apparent to a limited extent only.
Fig. 9.17. Section of the fracture surface

in the originally compressive zone from
the same specimen shown in Fig. 9.16
(from [130)).
The seemingly smooth parts of the cleavage surface deserve two further com-
ments. Figure 9 .18 shows that the material disintegration of the "smooth" surface
is not restricted to molecular dimensions but has a honeycomb-structure with cells
of 100 to 600 nm width. This is in accordance with the earlier statement on the size
of surface irregularities. It is also of interest to note that these dimensions are only a
few times larger than the radius of gyration of the chain molecules. The ridges be-
tween cells are highly deformed. As their width is comparable to that of the cells a
plastic deformation of about 100% is indicated. This is in view of the test temperature
quite remarkable. It is definitely caused by the fact that the comparatively long and
strong molecules are not highly oriented but have a coiled conformation correspond-
ing in all probability to the "solidification model" of semicrystalline polymers
(cf. 2 I C). Under these circumstances axial forces are only built up if a segment is
subjected to a sufficiently large shear displacement. Material separation thus neces-
sitates the deformation of several entangled molecular coils.
At room temperature molecular mobilities are increased and intermolecular
attraction is decreased. The extent of plastic deformation leading to material separa-
tion is, therefore, increased rendering some of the formerly brittle plastics tough
294
Fig. 9.18. Successive enlargements of the apparently smooth surface in the center of Fig. 9.16
(from [BDl).
(cf. 8 II A). In HDPE at room temperature no cleavage fracture (unstable crack

propagation) can be obtained by bending, it occurs only in notched specimens of
low aspect ratio or if the rate of loading is high (impact loading). In samples of HDPE
with extremely high molecular weight (Mw > 106 ) cleavage fracture does not occur
at all. Material separation must be enforced through large strains. Figure 8.28 gives
an impression of the very large plastic deformation of the fracture surface of a
notched bending specimen. The walls of the numerous rosettes and ridges are drawn
by several hundred per cent. The energy consumed in forming new surface area is
accordingly high. One can state that in short time loading of isotropic bulk polymers
chain molecules do not significantly contribute to the fracture surface energy through
the energies of bond breakage and elastic retraction. Large values of chain length and
chain strength reduce, however, the effect of natural or artificial flaws as stress con-
centrators due to the larger amount of plastic deformation necessary before material
separation occurs. Vice versa, at low molecular weights samples are increasingly more
brittle, fracture surfaces are smoother, secondary fracture events are rare or absent,
and the fracture surface energy accordingly small.
The above considerations apply to isotropic homogeneous materials, i.e. to
amorphous [61, 198, 200] and to semicrystailine polymers having no well defined
microstructure [130]. There the direction of the fracture path is more or less deter-
mined by the local stress field . In any event no traces of a microstructure became
apparent through the fracture surface morphology. This does not preclude the
existence of a certain globular microstructure (cf. 2 I C) which can be made visible
through ion-etching [132, 208]. Distinct fracture surface features are obtained, how-
ever, from polymers with a pronounced microstructure as introduced by extended-
chain crystallites or spherulites. There the material resistance depends strongly on
the relative orientation of the fracture plane with respect to the structural element.
Extended-chain crystallites have been studied by several authors (cf. 2 I C), most
recently by Gogolewski and Pennings et al. [201 - 203].
295
In Figures 9.l9-9.21 electron micrographs of their replicas from the fracture

surface of pressure crystallized polyamide 6 are reproduced [202]. The micrographs
reveal stacks of lamellae which have a thickness of up to 700 run. The authors con-
clude from their extensive IR, WAXS, and electron microscopical studies that the
lamellae represent extended chains. As they have proposed in Figure 9.22 a crack
may preferentially proceed either along the (010) planes (which contain the chain
ends as well as impurities rejected from the crystal growth front) or along the (002)
hydrogen bonded sheets of the lamellae. Both processes do not involve the breakage
of main chain bonds or of hydrogen bonds.
Fig. 9.1 9. Electron micrograph of a replica of a fracture surface of 6 polyamide crystallized at

295 °C for 48 h under a pressure of 6.5 kbar. The arrow shows the growth direction of the
spherulite (Courtesy S. Gogolewski (2021). (Reproduced by permission of the pUblishers,
IPC Business Press Ltd. ©).
Fig. 9.20 . Transmission electron micrograph of a replica of another part of the fracture surface
of the sample shown in Fig. 9.19 (Courtesy S. Gogolewski [2021). (Reproduced by permission
of the publishers, IPC Business Press Ltd. ©).
296
Fig. 9.21. Scanning electron micrograph of a

fracture surface of 6 polyamide crystallized
at 295 °C for 48 h under a pressure of
6.5 kbar. (Courtesy S. Gogolewski [202)).
(Reproduced by permission of the publishers,
With respect to extended chain crystallization under high pressure (p > 4 kbar)
the authors discuss two mechanisms [201-202]. Based on the observed change in
molecular weight distribution and decrease in molecular weight concurrent with
chain extension they suggest that the thermal treatment under high pressure makes
possible a transamidation reaction between -NH and -CO groups of broken chain
folds belonging to adjacent lamellae [201]. On the other hand the appearance of
lamellae with a thickness much smaller than the average chain length has led them
to the conclusion that fractionation is accompanying the pressure crystallization
[202].
The influence of a spherulitic structure of a semicrystalline polymer on its
mechanical properties has been implicitely and explicitely recognized for a long
time. Only a few references can be given here [183, 204-207]. Clark and Garber
HydrIKJen bonded sheets Crack
Fig. 9.22. Model proposed for fracture behavior

and surface morphology of chain-extended crystals
of polyamide (Courtesy S. Gogolewski (2021).
(Reproduced by permission of the publishers,
297
[205] review the effects of industrial processing on the morphology of crystalline

polymers. Patel et al. [206] and Andrews [207] are especially concerned with a
stress-strain analysis of spherulitic HDPE. They find a maximum of sample elastic
modulus of 1.2 GN m- 2 at a spherulite radius of 13 11m. Studies of the failure
morphology of HDPE in an active environment [204] revealed
that failure occurred almost entirely in a brittle mode
that failure within an individual spherulite is interlamellar, and
that the mode of crack propagation depended on the position of the spherulite
center with respect to the crack front (leading to interspherulitic fracture in the
case of poorly matching interfaces [204].
Extensive morphological studies of polypropylene fracture surfaces have been
carried out by Menges et al. [121,136] and quite recently by Friedrich [183]. Using
polypropylenes of different molecular weight and tacticity subjected to different
thermal treatment Friedrich [183] obtained samples of different microstructure:
I fine spherulites OS = 20 11m)
II coarse spherulites embedded in a matrix of finer spherulites
III fully grown coarse spherulites (is up to 500 11m).
The slow crack growth in CT specimens of morphology II is schematically
represented in Figure 9.23 a. Crazes were seen to develop ahead of the crack preferen-
Mafflx a
Fig. 9.23. Slow crack growth following craze formation in bulk polypropylene containing coarse
and fine spherulites (Courtesy K. Friedrich, Bochum).
(a) schematic representation, (b) crazes at the interface between coarse spherulites, (c) trans-
and interspherulitic crazing, (d) crazing of finely spherulitic matrix.
298
tially at the interfaces between coarse spherulites (Fig. 9 .23 b), within larger spheru-
lites (c) and at the boundaries of small ones (d). The behavior of a crack in the zone
of rapid crack propagation is illustrated in Figure 9.24. A scanning electron micro-
graph of the fracture surface reveals the structure-induced deviations of the fracture
path from a plane (Fig. 9.24a). The points 1 to 3 identify corresponding fracture
paths in Figure 9.24a, b, and c. These paths follow a spherulite boundary (1) or they
..-\---0)
b-=--=-~~. .~;;;~~---ffi
-=-- . Fig. 9.24. Rapid crack propagation in
~--- polypropylene. (Courtesy K. Friedrich,
Bochum).
(a) scanning electron micrograph of
fracture surface showing inter-(l) and
transspherulitic fracture (2 and 3),
(b) model representation, (c) optical
micrographs of sections transverse to
the fracture surface showing the same
phenomenon.
cut off respectively, a segment (2) , or a cone (3) . From his extensive morpho·
logical studies Friedrich [183] derived an order of the partial material resistance
values which decrease from that of the
separation of finely spherulitic matrix, which is the largest, to
straight paths through coarse spherulites and their center ,
radial and tangential planes in coarse spherulites, and
interfaces between coarse spherulites and fine matrix, to that of
, - polygonal interfaces between coarse spherulites.
299
The Kc values were found to decrease linearly with D, increase with decreasing
strain rate, and increase with molecular weight, the latter effect being especially
pronounced with the coarsely spherulitic morphology. Tentative absolute values of
R are determined for the separation of finely spherulitic matrix as 22 kJ m- 2 , and
for interspherulitic rupture as 4 kJ m -2 [183]. The first value is of the same order of
magnitude as that calculated by Kausch [11] for an oriented PP film (cf. further
below).
In the first part of this section, semicrystalline polymers (HOPE, PP, PA) have
been discussed. The fracture surface morphology of glassy polymers has received no
less attention in the literature. In fact many of the investigations of crazing [76-177]
use fractography as a means to elucidate the propagation and breakdown of crazes.
Fractographic studies of fracture processes concerning PS are especially reported in
refs. [106,115,132,150,155,169,191,194,199], concerningPMMA in [61, 66,
197,200], PVC in [198, 208], and PC in [196].
A feature common to practically all fracture surfaces of glassy polymers are the
remnants of craze layers. At low speeds of crack propagation craze breakdown will
generally occur in the center of the craze material leaving a more or less homogeneous
layer on each fracture surface [15, 50,150, 194, 199]. In PS at intermediate and
higher crack speeds decohesion at the craze/matrix interface becomes possible at
room temperature. Beahan et al. [150] have investigated this phenomenon more close-
ly; Figure 9.25 shows their micrograph of (the rather isolated) event of a crack which
Fig. 9.25. Transmission electron micrograph of the craze region ahead of a crack advancing from
left to right in polystyrene. (Courtesy D. Hull, Liverpool).
after having propagated at intermediate speeds was stopped in the craze region. The
micrograph permits recognition of the decohesion of highly strained craze material at
the interface and also the oscillation of this phenomenon between the opposite inter-
faces. The regular oscillation of the decohesion of craze material creates "mackerel
patterns" with typical band repeat distances of between 10 and 70 11m. Hull et al.
[150, 194, 199] discuss the dynamic conditions of their formation and the effect of
sample temperature. They observe that at temperatures below 235 K no mackerel
patterns are formed whereas the initiation of secondary fractures is greatly enhanced.
A more pronounced transition from slow to fast crack propagation is reported
for PMMA [61,66,197,200] but the surface features are roughly comparable to
those in PS. As observed by 0611 and Weidmann [61,66] at slow crack speeds
(li < 0.1 ms- l ), independent of molecular weight, a rather smooth fracture surface is
300
formed which even carries on accidental deviations from the plane in the form of
parallel markings (left hand side of Fig. 9.26). After the transition to rapid crack
propagation the fracture surface morphology changes completely. A pattern of ribs
or lines running approximately parallel to the crack front becomes visible. Although
length, direction, and spacing of the ribs are not very regular (Fig. 9.26, right hand
side) an average spacing of 90 f.1m has been measured [61]. At a somewhat higher
molecular weight (Mw = 163000) a spacing of about 220 f.1m is observed (Fig. 9.27,
right hand side). At very low molecular weight (Mw = 51000) rib spacing is 20 to
Fig. 9.26. Fracture surface of polymethylmethacrylate (Mw = 115000) showing the transition
from slow (Ii "" 0.1 ms- i ) to fast (a > 200 ms - i ) crack propagation; crack propagates from
left to right. (Courtesy W. Doll, IFKM Freiburg), Specimen thickness 4.75 mm.
Fig. 9.27. Fracture surface of polymethylmethacrylate (Mw = 163000) showing the transition
from a smooth surface (crack speed a= 300 ms - i ) to a coarse one (a > 400 ms- i ); crack
propagation is from left to right. (Courtesy W. Doll, IFKM Freiburg), Specimen thickness
3.75 mm.
301
24 11m [200]. In view of the symmetry of these markings on the two corresponding
fracture surfaces, rupture of the craze material in the center plane seems to be indi-
cated. This phenomenon is somewhat different in origin, therefore, from the mackerel
pattern but also related to the formation and breakdown of crazes. The mackerel
pattern was caused by the oscillating occurrence of craze/matrix decohesion. The rib
markings, however, are better interpreted as hesitation lines, also designated as stick-
slip, i.e. as the alternating development and breakdown of craze material [61,196,
200]. This is explained by referring to the craze model presented in Figure 9.14. It
may be assumed that at t = to the crack front just faces a fully developed craze whose
width and length (region C) comply with the Dugdale model (Eqs. 9.18 and 9.20).
The propagating stress field and the stresses which can be supported by the craze
zone are momentarily in equilibrium. At t = to + ~t, however, the craze is supposed
to open further and to support loads which are beyond the load carrying capability
of the molecular strands. Consequently the most advanced strands break thus initiat-
ing the rupture of neighboring strands as well. The catastrophic breakdown is slowed
down once the crack has advanced within region C so far as to reach little-drawn
craze material. After the slow-down of crack propagation redevelopment of the craze
begins which may be completed at, say, t = to + t 1 . Considering the average crack
speed (v = 400 ms- 1 ) and the rib spacing in Figure 9.27 (220 11m) one can calculate
that the time interval tl for catastrophic breakdown and redevelopment is 0.55I1S.
The cyclic crack propagation seems to be influenced, however, by additional mecha·
nisms which are not yet fully understood.
Thus the PMMA samples having the slightly higher molecular weight reveal at
crack speeds of 200 to 300 ms- 1 a rather smooth fracture surface containing no ribs
but a large number of parabolic markings (Fig. 9.27, left hand side). These markings
are seen in higher magnification in Figure 9.28, they derive from the coalescence of
primary and secondary crack fronts travelling in one craze zone but in slightly dif·
ferent planes. The fracture surface of polycarbonate - broken in tension at room
temperature - reveals a strikingly similar pattern [196]. In that case the distance
between the planes of primary and secondary cracks has been found to be 0.43 11m
at room temperature increasing to 0.75 11m at -196°C.
Doll [61] reports that PMMA samples with a higher molecular weight
(Mw ~ 490000) do not show ribs on the high-speed fracture surface (Fig. 9.29).
Secondary fractures (this time in different craze zones) may be initiated, however,
leading to parabolic markings in that surface region (Fig. 9.30). Kusy et al. [200]
find ribs and parabolas in the molecular weight range 92000 ~ Mv < 270000. A cer-
tain molecular coil size is (quite obviously) necessary to permit the transfer of
stresses sufficient for the initiation and temporary propagation of a secondary fracture
plane. So far no generally valid relation between material parameters, experimental
conditions, and morphological observations can be given.
In the referenced literature further details are discussed which may be obtained
from fractographic analysis, such as the effect of Wallner lines on the disposition of
the ribs [61,196,200], "crazeless fracture" in low molecular weight PS [155], the
retarded breakdown of crazes in fatigue (cf. III C), the ductile fracture ofPS at lower
loading rates and at temperatures close to Tg through the growth of one or more
diamond shaped cavities [169], the globular appearance of ion·etched PS craze matter
302
Fig. 9 .28. Higher magnification of the middle section of Fig. 9 .27 showing parabolic markings.
(Courtesy W. Doll, IFKM Freiburg) .
Fig. 9 .29 . Fracture surface of PMMA (Mw = 8000000) showing the transition from slow
(a '" 0.1 ms- 1) to fast (0 > 100 ms- 1) crack propagation; crack propagates from left to right.
(Courtesy W. Doll, IFKM Freiburg) , Specimen thickness 4.16 mm .
303
[132] and PVC surfaces [208], and craze-like features on the surface of broken phenol-
formaldeyhde samples [195].
In this section the morphology of fracture surfaces has been discussed, a mor-
phology which reflects the local modes of material separation. The microscopic size
of the structural elements to be broken or separated has been recognized: molecular
strands, fibrils or coils, ribs, crystalline lamellae, spherulites. In talking about their
ultimate response, however, macroscopic terms had been used: breakage, shear de-
formation, limits of plastic deformation, material resistance. For good reasons no
molecular criteria for material separation had been given. For individual molecules
such criteria exist: the temperature of thermal degradation and the stress or strain
which causes the chain to scission. For the mentioned structural elements no such
simple criteria exist. Perhaps one may mention the critical role of temperature in the
transition to rapid crack propagation [30, 50, 184-186, 197] and the constant value of
local strain €y in the extension direction (Fig. 9.31) which was found - independent
of crack length - to reach about 60% at the cracktip in biaxially oriented PETP film
[209]. One may also mention the critical concentration of chain end groups NIR
determined by microscopic infra-red spectroscopical investigation of the near-crack
zones of oriented PP film (Fig. 9.32) and of the fracture surface ofPE bulk material
[210]. Both materials are tough and strong. The given stress distribution in front of
the crack in the PP film permits calculation ofKc as a (r)yrirr = 8.3 ± 2 MN m- 3/ 2
and Gc as 30 ± 17 kJ m- 2 [11]. These values, in connection with Table 9.2, indicate
quite clearly that chain scission is accompanied by severe plastic deformation. The
Fig. 9.30. Parabolic markings from the surface of a crack travelling at 240 ms,.-l in PMMA
(M w = 1.2 · 106); crack propagates from left to right. (Courtesy W. Doll, IFKM Freiburg).
304
60,-------------------------------------~
10
distancE' to cracktlp
Fig. 9.31. Distribution of principal strain in front of a plane crack in biaxially drawn PETP fihn
(after (202)).
(Korsukov el at 1972 )
I r
~m 200
Fig. 9.32. Distribution of chain end concentration in front of a propagating plane crack at
t = 0.2 1" (+) and 0.751" (0) in polypropylene film (after [11, 210)).
305
possible role of chain scission during or after the application of large orientational
strains has been discussed in detail in Chapter 8.
B. Notched Tensile and Impact Fracture
In Chapter 8 tensile and impact behavior of unnotched specimens have been discussed.
The principal changes in behavior to be expected from the presence of a notch would
have to be due to the changed state and intensity of stress in the crack tip zone, to
the essential confinement of fracture development to a limited region, and to the
increase in rate of local deformation. An analytical description of these effects has
been attempted in terms of linear elastic fracture mechanics (9 I). Neglecting the
geometrical correction term the uniaxial brittle strength of a cracked thick plate is
derived from Eq. (9.10) as:
(9.29)
In this case the effect of a notch of length a is clearly indicated, therefore. In the
general case of rupture involving plastic deformation, however, one would have to use
Kc (or Gc ) and a corrected crack length a + f(rp) in Eq. (9.29). All these quantities
depend heavily on the degree of plastic deformation at the crack tip, which in turn is
influenced by crack length, notch tip radius, and loading conditions. In the case of
heavy plastic deformation the rupture stress of a specimen would depend little on
the initial crack geometry but strongly on the "strength" of the plastically deforming
matrix. Three questions are raised, therefore, which will be discussed in the following:
which notch sensitivity is actually observed with different polymers, when are geo-
metrical and when are material parameters of critical importance, and what is the
influence of chain length and mobility?
The notch sensitivity of polymeric materials has been comprehensively investi-
gated by Takano and Nielsen [211]. The authors define a notch sensitivity factor ks
of the yield or breaking strength of tensile bars as
ks = °F(O) A (a) (9.30)

0F(a) A(o)
with A being the (residual) sample cross-section. In the same way a notch sensitivity
factor k T of the energy to break (area under stress-strain curve) is defined.
The authors tested some 40 different materials at strain rates of 2.5 mm/min for
rigid materials and of 25 mm/min for soft elastomers employing six different crack
geometries. The "most severe" was a single notch of 3 mm depth (in a 12 mm wide
sample) and of 0.25 mm notch tip radius (notch s). The "least severe" was a double
notch of 3 mm depth and 1.5 mm notch tip radius (notch d). The authors measured
- where possible - the Izod impact strengths, the yield (OF) and breaking strengths
(Ob), and Young's modulus of notched and unnotched specimens.
From their measurements a sequence ofks(s) values can be established:
PMMA (4.71), Kraton ®-1l01 (3.68), SAN (2.71), PSU (2.36), EPDM rubber (2.12)
306
Ill. Molecular and Morphological Aspects of Crack Propagation
and various particle filled thermoplastics (2.14-1.33). The ks(d) values exhibit a
similar sequence:
PMMA (1.64), Kraton ® -11 01 « 1.69), SAN (1.62), and PPO (l.48). There are, how-
ever, quite a number of (ductile) materials with ks(s) values smaller than one,
notably HDPE (0.88), PTMT (0.96), PTFE (0.94), PA 6 (0.93), and Hytrel ® 4055.
Values below unity of ks(d) are more numerous than those of ks(s).
Notch sensitivity factors for energy, kT' show a somewhat different order. With
some rare exceptions (PP + 20% glass, PE + 40% glass) kT values are always larger
than unity. The factor kT(S) reaches values of 625 (semi-transparent PETP), 241 (PSU),
61 (PA 66 + 0.6% H 2 0), 48 (PP and PMMA), and 16 to 17 (mineral filled PA 66, PC).
For kT(d) the authors determine 81 (PA 66 + 0.6% H 2 0), 34 (PPO), 29 (semi-trans-
parent PETP), 19 (PP), and 14 (annealed HDPE and transparent PETP). All other kT
values measured are smaller than 13.
The data of Pagano and Nielsen [211] confirm the observation that the presence
of a crack not necessarily reduces the tensile strength of a sample or its energy to
break. The authors also note that in some cases (notably in HDPE, PTMT, PTFE,
PA 66 + 0.56% H 2 0) the "more severe" notch s is the less dangerous and has a smaller
kT value than notch d. The reason for this unexpected reversal is not clear but it may
be related to the volume of polymer under elevated stress in the notched region. This
volume at the tip of a blunt notch is larger than that at the tip of a sharp notch [211].
Regarding the experimental evidence (e.g. [58,64,211,214]) one may say that in
the case of a very sharp notch (Pa < 50 tlm) the onset of rapid crack propagation is
determined predominantly by notch length a and material fracture toughness Kc
according to Eq. (9.29). In a blunt notch specimen, however, the tangential stress
Oy at the root of the notch has to be considered the critical quantitiy [58, 64 ].If the
shape of a blunt notch is approximated by an ellipse the well known Neuber-Inglis
formula [4] can be used to calculate Oy:
(9.31 )
The stress concentration increases with decreasing P a whereas the apparent

critical energy release rate (Gc)blunt decreases. Considering also a plastic zone of size
rp Plati and Williams [64] obtain:
(G ) =G (l + Pa/ 2rp)3 (9.32)

c blunt c (1
+ Pa /)2
rp
Whenever the measurement of the energy to break a specimen in tension or

flexion is used to determine critical energy release rates, it must be recognized, that
this energy is the sum of a number of quite different terms. As outlined in 8 II A,
the energy loss An' which a pendulum incurs in striking and breaking a specimen, is
a measure of the elastically stored energy We, of the fracture surface energy Ws ' of
the kinetic energy Wkin of the broken pieces, and of otherwise dissipated energy.
The elastic energy to bend the specimen to a deflection 0 under load P amounts to:
We = ~ oP = ~ P 2 C, with C being the bending compliance. The energy to propagate

307
a crack through the ligament area is taken to be Ws = GcB(D-a). The problem of how
these two energy terms are related to the impact strength an has first been examined
independently and simultaneously by Brown and Marshall, Williams, and Turner [53];
it has recently been reviewed by Williams and Birch [69].
Crack propagation in a notched bending specimen becomes possible once the
release of stored elastic energy is equal to or larger than the material resistance:
p2 oC
G (a) = -
c 2 Boo
- > R(a). (9.33)
The crack will propagate in an unstable manner and without additional external
driving forces, i.e. at constant deflection, as long as the release of energy due to the
increase in specimen compliance C remains sufficient to supply the energy needed
for material separation, i.e. as long as
aGc(a) > oR(a) .

(9.34)
oa oa
If one makes use of Eq. (9.33) and of the condition of constant deflection 0,
which gives
oP/Poa = -oC/Coa. (9.35)
one obtains as condition for crack instability
C o2C/oa 2 BC oR/oa
(9.36)
"2 (oC/oa)2> I + p2 (oC/oa)2
IntrodUCing the "calibration factor" cp,
C
cp = oC/o(a/D) , (9.37)
taking R equal to Gc ' and rearranging Eq. (9.36) one arrives at a somewhat simpler
form of the instability condition:
ocp oR/Roa
(9.38)
o(a/D) <- 1- oC/Coa
If R is taken to be constant with a Eq. (9.38) is equal to the condition for crack
instability analyzed by Williams et al. [64,69].
The calibration factor cp(a/D) has been tabulated by Plati and Williams [64] for
the two most widely used impact tests, the three point bending (Charpy) test and the
cantilever bending (Izod) test (with aiD ranging from 0.04 to 0.60, 2 L/D from 4 to
12). If a bending specimen fails due to unstable crack propagation then its energy to
308
failure should principally correspond to the elastic energy stored at the point of onset
of rapid crack propagation, which can be expressed now as:
An ~ Wei = RBD1> (9.39)
A plot of An over BD1> should give, therefore, a straight line with slope R = Gc .
Testing nine different brittle and ductile polymers (from PS to PE) by the Charpy
and the Izod methods, Plati and Williams [64] were able to arrive at fairly unique
values of Gc (cf. Table 9.2). They point out, however, that in order to obtain linear
BD1> plots they had to consider an effective crack length consisting of the initial
crack length and the size of a suitably chosen plastic zone.
If the condition for unstable crack propagation is not fulfilled or maintained
throughout the available range of crack lengths then a continued supply of mechani·
cal energy is necessary to propagate the crack through the material. In that case the
failure energy An should be mainly proportional to the ligament area:
(9.40)
Corresponding plots of An over B (D-ao) for HiPS and ABS [53, 64, 69] revealed
this linear relationship.
The fact that even the rapid deformation of a glassy polymer under concentrated
stresses entails considerable local plastic deformation immediately suggests that the
molecular properties, which influence yielding and flow, are also effecting Gc and
thus the impact strength. The data compiled in Table 9·11 reveal this dependency of
Gc on temperature, rate of deformation, and molecular properties. A possible relation
between molecular relaxation processes and fracture surface energy of polymers has
been pointed out in many of the cited references (e.g. [14,19,22,24,25,54,63,
64,212-214]).
In 8 II A an account of the role of mechanical relaxation mechanisms in the
impact loading of unnotched specimens has been given. The reasons for an expected
positive correlation (and for the also observed deviations) between impact strength
and magnitude of mechanical losses have been indicated there. At this point the
molecular aspects of notched impact strength are to be investigated. Sauer [213] and
Vincent [214] have reviewed the impact and stress-relaxation data from a large
number of publications including their own extensive works and those referenced in
8 11 A. They arrive at various general conclusions which may be listed in the following:
The impact resistance rating of a polymer is generally determined by the storage
component E' of its dynamic modulus. For some twenty different polymers tested,
65% of the data complied with the following correlation [214]:
Impact resistance rating Modulus
- brittle E' > 4.49 GN m- 2
- brittle if bluntly notched E' ~ 3 GN m- 2
- brittle if sharply notched E' ~ 2.2 GN m- 2
- tough but crack propagating E' < 1.5 GN m- 2
Deviations from this scheme were either related to molecular structure (bulky side
groups ofPMMA), to mechanical loss peaks (PTFE, HDPE, PC, PPO), to mor·
309
phology (injection molded PP), or to heterogeneous reinforcements (short glass

fibers, particulate fillers).
- Polymers with high impact strength at room temperature also have a significant
low temperature loss peak (tan [) ~ 5 . 10-2 or higher). Polymers in this class
include, PE, PC, PB, PTFE, PCTFE, POM, and the polyamides.
- Polymers with low impact strength at room temperature usually have no signifi-
cant low temperature loss peaks in the 100 K to 300 K range (PS, PMMA).
- Impact strengths have been measured as a function of temperature in several
polymers and it has been shown that in many cases one gets a rise in brittle impact
strength in the vicinity of the low temperature 'Y-relaxation (PTFE, PE, POM).
For example, in PTFE at about 210 K, in PE at 150 K. Good correlation has also
been observed between the magnitude of the low temperature 'Y-loss peak in poly-
sulfone and impact resistance for a series of specimens containing various amounts
of an antiplasticizer.
- An increase in molecular weight will generally affect the impact strength through
the increase in Gc ; this is most notably observed in HDPE (cf. 8 II A).
- In high impact polystyrene (HiPS) the normally brittle PS matrix has been blended
with a suitable rubber-type component such as a styrene-butadiene copolymer
which has a glass transition well below room temperature. The impact strength
increases in proportion to the amount of the second component added. In this
case of a two-phase polymer, however, impact resistance is increased through the
initiation of a myriad of fine crazes and not through the generally increased
extensibility of a homogeneously deforming matrix.
C. Fatigue Cracks
The characteristics of fatigue loading, i.e. of the repeated application of varying stress
or strain amplitudes have been introduced in detail in Section 8 II C. In that discussion
one point was left open: that of the mechanism of fatigue crack propagation. A com-
prehensive elaboration of this subject has quite recently been given by Hertzberg in
his book on "Deformation and Fracture Mechanics of Engineering Materials" [3].
The reader is referred to this work or to the review articles by Plumb ridge [217] and
Manson and Hertzberg [218] for a detailed discussion of the different stages of
fatigue crack growth, of the distinct characteristics of fatigue fracture surfaces, of
the various theoretical approaches to derive crack growth-rate equations, and of the
S-N curves of a wide variety of homogeneous and fiber reinforced polymers - and
of metals, for that matter [3, 217, 218]. At this point only some recent observations
will be discussed which seem to be directly related to the chain-like nature of macro-
molecules [173,178,191,215-220].
As to be expected, the growth of a crack under constant or increasing load and
fatigue crack propagation have quite a few features in common. Thus both the static
and the fatigue fracture surfaces reveal a slow growth and a rapid crack propagation
zone. The stress intensity factors reached in both cases at the transition from stable
to unstable crack propagation correlate very well [218]. The rate of stable fatigue
crack growth is empirically expressed by most authors in form of
310
(9.41)
with A and n material parameters and AK the applied stress intensity factor range.
This is in complete analogy to the expression for the rate of crack propagation under
static loading (Eq. 9.22). Marshall et al. [133] and Radon et al. [219] substitute AK
by the difference of K2 and K3 terms. Andrews et al. [215] employ in their analysis
of the fatigue of PE the energy release rate Gc (instead of AK) as independent param-
eter. Such approaches are not significantly different because of the relation between
K and G and of the form of Eq. (9.41). In their detailed reports [3, 218] Hertz-
berg and Manson analyse the various theoretical approaches, furnish fatigue crack
growth data for some 20 materials at different temperatures and frequencies, and
discuss the effect of environmental and material parameters.
Skibo, Hertzberg, and Manson [191] studied fatigue crack growth characteristics
in polystyrene as a function of stress intensity factor range and cyclic frequency.
Precracked single edge notched and compact-tension type specimens made from
commercially available polystyrene sheet (mol. wt. = 2.7 . 105 ) were cycled under
constant load at frequencies of 0.1, 1, 10 and 100 Hz, producing growth rates rang-
ing from 4 . 10- 7 to 4· 10- 3 cm/cycle. For a given stress intensity level, fatigue
crack growth rates were found to decrease with increasing frequency, the effect being
strongest at high stress intensity values. The variable frequency sensitivity of this
polymer over the test range studied was explained in terms of a variable creep com-
ponent. The macroscopic appearance of the fracture surface showed two distinct
regions. At low stress intensity values, a highly reflective, mirror-like surface was
observed which transformed to a rougher, cloudy surface structure with increasing
stress intensity level. Raising the test frequency shifted the transition between these
areas to higher values of stress intensity. The microscopic appearance of the mirror
region revealed evidence of crack propagation through a single craze while the appear-
ance of the rough region indicated crack growth through many crazes, all nominally
normal to the applied stress axis. Electron fractographic examination of the mirror
region revealed many parallel bands perpendicular to the direction of crack growth,
each formed by a discontinuous crack growth process as a result of many fatigue
cycles. The size of these bands was found to be consistent with the dimension of the
crack tip plastic zone as computed by the Dugdale model. At high stress intensity
levels a new set of parallel markings was found in the cloudy region which corre-
sponded to the incremental crack extension for an individual loading cycle [191].
In these experiments two chain-related phenomena have been observed: the dis-
continuous growth of a fatigue crack and the frequency sensitivity of the growth
rate. The discontinuous growth of a fatigue crack had been noted earlier for other
polymers as well: PVC [216], PMMA, PC, and PSU [191]. Elinck et al. [216] report
that, depending on frequency, between 130 and 370 load cycles were necessary to
form one "arrest line" in PVC; Skibo et al. [191] found 600 to 1400 cycles for PS.
The width of these growth bands corresponded well to the plastic zone size rp cal-
culated from the Dugdale model. Based on these findings Skibo et al. [191 ] propose
as a possible crack growth mechanism the gradual development of a craze of width
rp and its subsequent and rapid breakdown. Such a mechanism would also be sup-
311
ported by the model of craze formation given in G II C: once the critical and fre-
quency-dependent fibril extension is reached at the crack tip cooperative breakdown
of the fibrils occurs and the crack propagates from D/C almost to B (Fig. 9.14).
The frequency-dependent rate of disentanglement of the molecular coils in the
fatigued fibrils seems to account for part of the frequency effect on growth rate.
In addition hysteresis heating occurs in the straL'led craze material. Both effects
combine to give rise to a distinct frequency sensitivity of A for a variety of materials
such as PC and PMMA [219, 220], and PPO,PVC,PA 66,PC,PVDF, and PSU[220].
As has been noted by Skibo et al. [220] the frequency sensitivity varies with tem-
perature. It reaches a maximum at that temperature where the external (fatigue)
frequency corresponds to the internal segmental jump frequency (of the (3-relaxation
process).
The effects of strain amplitude, sample molecular weight, environment, and
surface coating on the fatigue properties of PE and PS have been studied by Sauer
et al. [173, 178]. Their results have been discussed in 8 II C. It is especially note-
worthy that the initiation of a fatigue crack can be retarded by a decade or more
through the application of a compatible, viscous coating. A 600 molecular weight
PS oligomer served this purpose for both polished and unpolished surfaces of cylin-
drical PS specimens [178].
The relation between microstructure of LDPE and fatigue crack growth was
investigated by Andrews et al. [215]. They found growth rates according to Eq. 9.41
in two regions: that of the initial brittle transpherulitic crack propagation and that
of the ductile crack propagation at higher K values. Between these two regions a
distinct transition region was found where da/dN F depended little on ~K.
D. Mechano-Chemistry
In a broader sense mechano-chemistry comprises all aspects of the stress-induced

scission of chain molecules. In a narrower sense, however, one speaks of mechano-
chemical methods if one refers to the intentional mechanical degradation of (solid)
polymers. The objectives in the latter cases are the comminution or softening of the
material or the production of large and highly reactive surfaces as a means to initiate
permanent chemical bonding between different polymers. In Table 9.5 an overview
on methods and processes is given which may result in a mechanical degradation of
chain molecules. The objectives of these processes are indicated with respect to the
deformation mechanism. It should be pointed out that in the mechano-chemical
methods, in the narrower sense, the degrading solids are subjected to ill defined, com-
plex states of stress leading to a deformation which always includes simultaneously
the yielding and flowing of material and the breakage of chains. Table 9.5 lists
the deformation mechanisms which are the most important in view of the corre-
sponding objective. Reference is made to those chapters and sections of this book
where the particular deformation mechanisms are treated.
In recent years a number of comprehensive review articles on the mechano-
chemistry of polymers have appeared [221-226]. A competent two-volume hand-
book by Porter and Casale on this subject [227] has just become available. The
312
Table 9.5. Mechano-chemistry of Solid Polymers: Methods and Objectives
Method, Process Objective Deformation Mechanism Reference to

Chapter Section
Grinding, sawing, cutting comminution compressive yielding, S.Il. D

Ball or vibromilling comminution I elastic compressive S.Il. E
Impact loading shock absorption I and/or tensile deforma-
tion
Mastication softening stretching and breakage 5, S.Il.E
(of elastomers) of chain molecules
Extrusion, injection, forming shear flow S.Il. D
drawing
Mixing, stirring dispersion of additives frictional interaction,
compressive yielding
Synthesis through grafting, copolymeriza- chain scission and free 6.IV. A, 7.III.
mechanical blending tion of incompatible or radical reactions
mechanically different
polymers
reader is referred to these references [222-227] for any extended information on

the effect of experimental variables (type of equipment, temperature, rate of mechani-
cal working, atmosphere) or polymer characteristics (chemical structure, initial molecu-
lar weight) on the resulting material properties.
Within this monograph special consideration is given to the loading and breakage
of molecular chains. With regard to mechano-chemical processes such as comminution
or mechanical synthesis one has to study, therefore,
the response of the polymer matrix to mechanical working
the consequences of chain scission: reduction of molecular weight and radical
formation
the results of radical reactions, e.g. oxidative degradation, cross-linking, copoly-
merization.
The property changes resulting from a mechanical treatment of polymeric
materials have been known and utilized for a long time. Thus knowledge of the fact
that mechanical working of natural rubber leads to its softening already dates back
120 years [225]. But it was only after the introduction of the concept of macro-
molecules some fifty years ago that this effect of mechanical treatment was related
to the rupture of molecular chains. The fact that the intensive milling of a hetero-
phase mixture (of rubber and maleic acid anhydride) can induce a chemical reaction
between the components was first observed in 1941 [224]. Systematic studies of the
nature of these chemical reactions and particularly of the role of the formed free
radicals began about ten years later [224-225].
The response of polymer matrices and of chain molecules to mechanical working
have been intensively discussed in the previous chapters. The ESR investigations of
ground polymers (6.1V A and 7.1 C 3) have shown that in a comminution process
(e.g. milling) chain molecules are broken in large numbers and as a consequence of
the formation of new surface during the breaking-up of the polymer particles. The
313
chain ruptures have an immediate and notable effect on the sample molecular weight
M [221-338]. In Figure 9.33 M is plotted as a function of particle diameter D. The
linear sections may be represented by a potential law
(9.42)
where A and n are material-related constants. Curves a and b in Figure 9.33 reveal
that chain scission is more effective in polystyrene (n = 0.76) than in PMMA
(n = 0.50). In both cases co~inution was carried out [228] in the presence of
radical scavengers (N0 2 ) to prevent chain recombination or cross-linking. Comminu-
tion in air seems to have permitted cross-linking reactions to such an extent that the
average molecular weight of a particle size fraction did not change with D (curve c).
The opposite effect, decrease of M without increase of specific surface was noted by
Komissarov et al. [222] in the dispersion of PA and PETP fibers without cooling.
I02~----------~-----------,
Fig. 9.33. Molecular weight of

comminution products as a
function of particle size
(after (200) from (2011):
(a) polystyrene in N02.
(b) polymethylmethacrylate
block copolymer in N02.
(c) polystyrene in air.
It is a general observation that in continuous mechanical working D (and M) do

not decrease indefinitely but reach certain limiting values. These limiting average M
values are of the order of the entanglement molecular weight Me. This does not
preclude the possibility that in mechanical degradation fractions of low molecular
weight and even volatile components can be produced [221-225].
The effect of environmental temperature on the rate of degradation has also been
widely studied [221-227]. In view of the composite nature of the degradation process
no simple rate equations can be expected. It becomes clear from Eqs. (5.41) and (7.3)
that the softening of the matrix (decrease of ~ 0) and the reduced effective bond
strength U(T) partly balance each other. According to an extensive review by Casale
[226] the temperature effect on matrix rigidity seems to be the dominant one. The
slower relaxation times at lower temperatures lead to an increase in mechanical degra-
dation with decreasing temperature (negative temperature coefficient of the overall
mechano-chemical reaction).
314
1. A. A. Griffith: Phil. Trans. Roy. Soc. (London) A 221,163-198 (1920).

2. Mechanics of Fracture, G. C. Sih, ed., Vol. I-V, Leyden: Noordhoff International Publishing.
1975-1977.
3. R. W. Hertzberg: Deformation and Fracture Mechanics of Engineering Materials, New York:
John Wiley 1976.
4. J. G. Williams: Stress analysis of polymers, London: Longman 1973.
5. E. H. Andrews: Fracture in Polymers, Edinburgh: Oliver + Boyd, 1968.
6. F. Kerkhof: Kolloid-Z. u. Z. Poly mere 251,545 (1973).
7. D. S. J. Dugdale: Mech. Phys. Solids 8, 100 (1960).
8. ASTM Book of Standards, Part 31: Physical and Mechanical Testing of Metals (1969).
H. J. G. Blauel, J. F. Kalthoff, E. Sommer: Materialpriifung 12,69 (1970).
9. J. D. Eshelby: Proc. Royal Soc. A241, 376-396 (1957).
10. G. R. Irwin: Handbuch der Physik 6, Berlin: Springer Verlag 1958,559.
11. H. H. Kausch: Kunststoffe 66/9, 538-544 (1976).
12. G. P. Marshall, L. H. Coutts, J. G. Williams: Materials Sci. 9, 1409 (1974).
13. P. W. Beamont, R. J. Young: J. Materials Sci. 10, 1334-1342 (1975).
14. F. A. Johnson, J. C. Radon: J. Polymer Sci., Polymer Chern. Ed. 11, 1995-2020 (1973).
15. G. W. Weidmann, W. Doll: Colloid and Polymer Sci. 254, 205-214 (1976).
16. J. P. Berry, in: Fracture Processes in Polymeric Solids, B. Rosen, ed., New York: Interscience
1964, Chap. II.
17. L. J. Broutman, F. J. McGarry: J. appl. Polymer Sci. 9, 589-626 (1965).
18. M. F. Kanninen, A. R. Rosenfield, R. G. Hoagland, in: Deformation and Fracture of High
Polymers, Kausch, Hassell, Jaffee Eds., New York: Plenum Press 1974,471-484.
19. M. Parvin, J. G. Williams: J. Materials Sci. la, 1883-1888 (1975).
20. Y. W. Mai: Int. J. Fracture 9, 349 (1973).
21. R. P. Kambour, S. Miller: General Electric Report No 77CRD009, March 1977.
R. P. Kambour, A. S. Holik, S. Miller: General Electric Report No 77CRDI25, June 1977.
22. S. Arad, J. C. Radon, L. E. Culver: J. appl. Polymer Sci. 17, 1467-1478 (1973).
23. W. Retting: Materialpriifung 5, 101 (1963).
24. W. Retting: Kolloid-Z. Z. Polymere 210,54 (1966).
25. H. W. Greensmith, L. Mullins, A. G. Thomas: Trans. Soc. Rheol. 4, 179 (1960).
26. G. J. Lake, P. B. Lindley, A. G. Thomas: Paper 43 (Session IV) Int. Conf. on Fracture 1969,
Brighton.
27. A. G. Thomas, in: Deformation and Fracture of High Polymers, New York: Plenum Press
1973,467-470.
28. A. Ahagon, A. N. Gent: J. Polymer Sci., Polymer Physics Ed. 13, 1903-1911 (1975).
29. H. K. Mueller: Ph. D. Thesis, California Inst. of Technology, Pasadena, California, June
1968.
30. W. Doll: Energieumsetzung in Wiirme beim kritischen Bruchbeginn in PMMA. Report of the
Institut fiir Festkorpermechanik, Freiburg, 1978.
31. D. W. van Krevelen: Properties of Polymers, Amsterdam: Elsevier Publ. Co. 1972, Table 6.6.
Polymer Handbook, Brandrup, J., Immergut, E. H., J. Wiley + Sons, New York 1975.
32. J. J. Benbow, F. C. Roesler: Proc. Phys. Soc. B 70, 201-211 (1957).
33. J. P. Berry: J. Polymer Sci. 50, 107-115 (1961).
34. J. P. Berry: J. Polymer Sci. 50, 313-321 (1961).
35. N. L. Svensson: Proc. Phys. Soc. 77, 876-884 (1961).
36. J. J. Benbow: Proc. Phys. Soc. 78, 970 (1961).
37. J. P. Berry: J. appl. Physics 34,62 (1963).
38. A. van den Boogaert: The Physical Basis of Yield and Fracture, Oxford, 1966, pg. 167.
39. P. I. Vincent, K. V. Gotham: Nature 210, 1254 (1966).
40. J. P. Berry: J. appl. Physics 34, 62-68 (1968).
41. J. O. Outwater, D. J. Gerry: J. Adhesion 1, 290 (1969).
42. S. Mostovoy, P. B. Crosely, E. J. Ripling: J. Materials, 2, 661 (1967).
315
43. G. P. Marshall, L. E. Culver, J. G. Williams: Polymer 37,75-81 (1969).

44. R. Griffith, D. G. Holloway: J. Materials Sci. 5, 302-307 (1970).
45. J. W. Curtis: J. of Physics 3, 1413-1422 (1970).
46. L. E. Miller, K. E. Puttick, J. G. Rider: J. Polymer Sci. 33/C, 13-22 (1971).
47. B. Harris, P. W. R. Beaumont, E. Moncunill de Ferran: J. Materials Sci. 6, 238-251 (1971).
48. A. R. Rosenfield, P. N. Mincer: Polymer Sci. Symposium 32, New York: John Wiley (1971),
283-296.
49. J. S. Foot, I. M. Ward: J. Materials Sci. 7, 367-387 (1972).
50. W. Doll: Kolloid-Z. u. Z. Polymere 250, 1066-1073 (1972).
51. G. P. Marshall, L. E. Culver, J. G. Wi11liams: Int. J. Fracture 9,295 (1973).
52. J. G. Wi11iams, G. P. Marshall, in: Deformation and Fracture of High Polymers, H. H. Kausch,
J. A. Hassell, R. I. Jaffee Eds., New York: Plenum Press 1973,557.
53. H. R. Brown: J. Materials Sci. 8, 941-948 (1973); G. P. Marshall, J. G. Williams, C. E. Turner,
ibid., 949-956 (1973).
54. F. A. Johnson, A. P. Glover, J. C. Radon: Proc. 1973 Symposium on Mechanical Behavior
of Materials, 141-148, The Society of Materials Sci., Japan (1974).
55. B. E. Nelson: J. Colloid and Interface Sci. 47/3, 595-599 (1974).
56. P. W. R. Beaumont: J. Adhesion 6, 107-137 (1974).
57. G. Goldbach: Kunststoffe 64/9, 475-4,81 (1974).
58. R. A. W. Fraser, I. M. Ward: J. Materials Sci. 9,1624-1630 (1974).
59. A. D. S. Diggwa: Polymer 15,101 (1974).
60. W. Doll: Int. J. Fracture 11, 184-186 (1975).
61. W. Doll: J. Materials Sci. 10, 935-942 (1975).
62. R. Ravetti, W. W. Gerberich, T. E. Hutchinson: J. Materials Sci. 10, 1441-1448 (1975).
63. F. A. Johnson, J. C. Radon: J. Polymer Sci., Polymer Chemistry Ed. 13, 495-516 (1975).
64. E. Plati, J. G. Williams: Polymer 16,915-920 (1975). E. Plati, J. G. Williams: Polymer Engng.
and Sci. IS, 470-477 (1975).
,65. R. P. Kusy, M. J. Katz: J. Materials Sci. 11, 1475-1486 (1976). R. P. Kusy, D. T. Turner:
Polymer 17, 161-166 (1976).
66. W. Doll, G. W. Weidmann: J. Materials Sci., Letters 11,2348-2350 (1976).
67. W. Doll: Int. J. Fracture 12/4,595-605 (1976).
68. Y. W. Mai, J. G. Williams: J. Materials Sci. 12, 1376-1382 (1977).
69. J. G. Wi11iams, M. W. Birch: Fracture 1977, Vol. I, ICF4, Waterloo Canada, June 19-24.
1977, pg. 501-528.
70. W. Doll, E. Gaube: unpublished results.
71. N. J. Mills, N. Walker: Polymer 17, 335-344 (1976).
72. P. S. Leevers, J. C. Radon, L. E. Culver: Polymer 17, 627 -632 (1976).
73. R. A. Gledhill, A. J. Kinloch: Polymer 17, 727-731 (1976).
74. G. P. Morgan,1. M. Ward: Polymer 18,87-91 (1977).
75. S. Yamini, R. J. Young: Polymer 18,1075-1084 (1977).
76. R. P. Kambour: J. Polymer Sci. D (Reviews) 7, 1-154 (1973).
77. S. Rabinowitz, P. Beardmore: CRC Critical Reviews I, 1 (1973).
78. C. C. Hsiao, J. A. Sauer: J. of Appl. Physics 21/11,1071-1083 (1950).
J. A. Sauer, C. C. Hsiao: Trans. ASME 75, 895 (1953).
79. J. F. Rudd: J. Polymer Sci. BI, 1 (1963).
80. H. A. Stuart, D. Jeschke, G. Markowski: Materialpriifung 6, 77 (1964).
81. A. C. Knight; J. Polymer Sci. A, 3, 1845 (1965).
82. G. Rehage, G. Goldbach: Die Angewandte Makromolekulare Chernie 1,125-149 (1967).
83. R. P. Karnbour, R. W. Kopp: General Electric, Report No. 67-C-374 (1967).
84. L. C. Cessna, Jr., S. S. Sternstein: Fundamental Phenomena in the Materials Sci. 4, (plenum
Press), 45-79 (1967).
85. R. P. Kambour: Applied Polymer Symposia 7, 215-235 (1968).
86. S. S. Sternstein, L. Ongchin, A. Silverman: Applied Polymer Symposia 7, 175-199 (1968).
87. R. P. Karnbour, A. S. Holik: General Electric, Report No. 69-C-022 (1968).
88. G. M. Bartenev: Vysokomol. soyed. AU/IO, 2341-2347 (1969). Polymer Sci. USSR 11/10,
2663-2671 (1970).
316
89. R. N. Haward, B. M. Murphy, E. F. T. White: Paper 45 (Session IV), Fracture 1969, Brighton,
45/1-45/13.
90. P. Dunn, G. F. Sansom: J. appl. Polymer Sci. 13,1641-1655 (1969).
91. P. Dunn, G. F. Sansom: J. appl. Polymer ScLI3, 1657-1672 (1969).
92. P. Dunn, G. F. Sansom: J. appl. Polymer ScLI3, 1673-1688 (1969).
93. P. Dunn, G. F. Sansom: J. appL Polymer Sci. 14, 1799-1806 (1970).
94. R. S. Moore, C. Gieniewski: J. appl. Polymer Sci. 14, 2889-2904 (1970).
95. K. C. Rusch, R. H. Beck, Jr.: J. MacromoL ScL-Phys. B 3(3) 365 (1969), B 4(3), 261 (1970).
96. A. N. Gent: J. Materials Sci. 5, 925 (1970).
97. L. Morbitzer, R. Holm, U. Brenneisen, K. Heidenreich, H. Rohr: Kunststoffe 60/11, 861-866
(1970).
98. J. Murray, D. Hull: Polymer Letters 8, 159-163 (1970), J. Murray, D. Hull: J. Polymer Sci.
A-2,8, 1521-1543 (1970).
99. Yu. M. Malinskii, V. V. Prokopenko, N. A. Ivanova, V. A. Kargin: Mekh. Polimerov (USSR)
6/2,271-275 (1970), Polymer Mech. (USA) 6/2, 240-244 (1973).
100. Yu. M. Malinskii, V. V. Prokopenko, N. A. Ivanova: Mekh. Polimerov (USSR) 6/3, 445-448
(1970), Polymer Mech. (USA) 6/3, 382-384 (1970).
101. F. Fischer: Z. f. Werkstofftechnik 1/2,74-83 (1970).
102. G. P. Marshall, L. E. Culver, J. G. Williams: Proc. Roy. Soc. Lond. A319, 165-187 (1970).
103. G. Menges, H. Schmidt, H. Berg: Kunststoffe 60/11, 868-872 (1970).
104. G. Menges, H. Schmidt: Plastics & Polymers 38, 13-19 (1970).
105. T. T. Wang, M. Matsuo, T. K. Kwei: J. appL Physics 42/11,4188-4196 (1971).
106. P. Beahan, M. Bevis, D. Hull: PhiL Magazine 24/12,1267-1279 (1971).
107. N. Fujimoto, M. Takayanagi, Y. Yamaguchi: Proceeding of Int. Conf. Mech. Behavior of
Materials, Kyoto, 1971,572.
108. R. N. Haward, B. M. Murphy, E. F. T. White: J. Polymer Sci. A-2/9, 801-814 (1971).
109. B. J. Macnulty: J. Materials Sci. 6, 1070-1075 (1971).
110. J. Pohrt: J. Macromol. Sci.-Phys. B5(2), 299-316 (1971).
111. J. Pohrt: Gummi Asbest Kunststoffe 24,594-606, and 700-713 (1971). K. V. Gotham:
Plastics and Polymers 40,277-282 (1972). .
112. Yu. S. Lipatov, F. B. Fabulyak: J. Appl. Polymer Sci. 16,2131 (1972).
113. H. Schmidt: Kunststoff-Rundschau 19/1,1-7 (1972). H. Schmidt: Kunststoff-Rundschau
19/2,3,56-65 (1972).
114. C. B. Bucknall, D. Clayton: J. Materials Sci. 7,202-210 (1972).
115. P. Beahan, M. Bevis, D. Hull: J. Materials Sci. 8, 162-168 (1972).
116. N. Brown, M. F. Parrish: Polymer Letters Ed. la, 777-779 (1972).
117. L. Spenadel: J. Appl. Polymer ScLI6, 2375-2386 (1972).
118.1. Narisawa, T. Kondo: Int. J. of Fracture Mechanics 8/4, 435-440 (1972).
119.1. Narisawa: J. Polymer Sci. Part A-2/IO, 1789-1797 (1972).
120. J. Hoare, D. Hull: Phil. Magazine 26/2, 443-455 (1972).
121. G. Menges, E. Alf: Kunststoffe 62/4,259-267 (1972).
122. H. E. Driesen: gwf-gas/erdgas 113/6,265-270 (1972).
123. P. 1. Vincent, S. Raha: Polymer 13/6, 283-287 (1972).
124. E. H. Andrews, L. Bevan: Polymer 13/7,337-346 (1972).
125. R. P. Kambour, E. E. Romagosa, C. L. Gruner: Macromolecules 5/4, 335-340 (1972).
126. E. H. Andrews, G. M. Levy, J. Willis: J. Materials Sci. 8,1000-1008 (1973).
127. L Narisawa, T. Kondo: J. Polymer Sci. - Poly. Phys. Ed. -11,223-232 (1973).
128. R. P. Kambour, C. L. Gruner, E. E. Romagosa: J. Polymer Sci. - Poly. Phys. Ed. -11,
1879-1890 (1973).
129. N. Brown: J. Polymer Sci. - Poly. Phys. Ed. -11, 2099-2111 (1973).
130. E. Gaube, H. H. Kausch: Kunststoffe 63/6,391-397 (1973).
131. J. Lilley, D. G. Holloway: Phil. Mag. (8) 28/1,215-220 (1973).
132. K. P. Grosskurth: Kautschuk u. Gummi-Kunststoffe 26/2, 43-45 (1973).
133. G. P. Marshall, J. G. Williams: J. Appl. Polymer Sci. 17, 987-1005 (1973).
134. S. Fischer, N. Brown: J. Appl. Phys. 44/10, 4322-4327 (1973).
317
135. H. G. Olf, A. Peterlin: Polymer 14,78-79 (1973).

136. G. Menges: Kunststoffe 63,95-100 a. 173-177 (1973).
137. R. N. Haward, D. R. J. Owen: J. Materials Sci. 8, 1135-1144 (1973).
138. H. G. Olf, A. Peterlin: Macromolecules 6,470-472 (1973). H. G. Olf, A. Peterlin: J. Colloid
and Interface Sci. 47/3, 628-634 (1973).
139. S. S. Sternstein, F. A. Myers: J. Macromol. Sci.-Phys. 8/3-4,539-571 (1973).
140. J. R. Kastelic, E. Baer: J. MacromoL Sci.-Phys. 7/4,679-703 (1973).
141. L. Camwell, D. Hull: Phil Magazine 27/5,1135-1150 (1973).
142. R. J. Oxborough, P. B. Bowden: Phil. Magazine 28/3,547-559 (1973).
143. K. J. Takahashi: Polymer ScL - Poly. Phys. Ed. - 12, 1697-1705 (1974).
144. M. Kitagawa, K. Motomura: J. Polymer Sci. - Poly. Phys. Ed. -12,1979-1991 (1974).
145. H. G. Olf, A. Peterlin: J. Polymer Sci. - Poly. Phys. Ed. -12, 2209-2251 (1974).
146. J. F. Feners, B. F. Kee: J. AppL Polymer Sci. 18, 2355-2365 (1974).
147. E. H. Andrews, G. M. Levy: Polymer 15/9, 599-607 (1974).
148. T. L. Peterson, D. G. Ast, E. J. Kramer: J. Appl. Phys. 45/10, 4220-4228 (1974).
149. K. P. Grosskurth: Kautschuk u. Gummi - Kunststoffe 27/8,324-328 (1974).
150. P. Beahan, M. Bevis, D. Hull: Proc. R. Soc. Lond. A 343,525-535 (1975).
151. G. W. Weidmann, J. G. Williams: Polymer 16/12, 921-924 (1975).
152. A. S. Argon: Pure and AppL Chemistry 43/1-2,247-272 (1975).
153. D. Hull, L. Hoarse: Plastics and Rubber: Materials and Appl. 5, 65-73 (1976).
154. P. Filzek, R. Siiselbeck, W. Wicke: Kunststoffe 66/1,38 (1976).
155. M. J. Doyle: J. Polymer Sci., Polymer Phys. Ed., 13, 127-135 (1975).
156. N. Verheulpen-Heymans, J. C. Bauwens: J. Materials Sci.H, 1-6 (1976). N. Verheulpen-
Heymans, J. C. Bauwens: J. Materials Sci.H, 7-16 (1976).
157. N. Verheulpen-Heymans: J. Polymer Sci.: Polymer Phys. Ed., 14, 93-99 (1976).
158. Y. Imai, N. Brown: J. Materials Sci. 11, 417 -424 (1976). Y. Imai, N. Brown: J. Materials
Sci.H, 425-433 (1976).
159. H. G. Krenz, D. G. Ast, E. J. Kramer: J. Materials Sci.H, 2198-2210 (1976).
160. H. G. Krenz, E. J. Kramer, D. G. Ast: J. Materials ScL II, 2211-2221 (1976).
161. D. L. G. Lainchbury, M. Bevis: J. Materials Sci.H, 2222-2234 (1976).
162. D. L. G. Lainchbury, M. Bevis: J. Materials ScLH, 2235-2241 (1976).
163. S. Sikka: PhD Dissertation, University of Utah, Salt Lake City, Utah (1976).
164. I. D. Graham, J. G. Williams, E. L. Zichy: Polymer 17/5, 439-442 (1976).
165. A. S. Argon, M. I. Bessonov: Reports of Research in Mech. Process. of Polymers 12, MIT
(1976).
166. A. S. Argon, J. G. Hannoosh, M. M. Salama: Reports of Research in Mech. Process. of Polymer
17, MIT (1977); Fracture 1977, VoL 1, Waterloo, Can., pg. 445. A. S. Argon, J. G. Hannoosh:
Reports of Research in Mech. Process. of Polymers 21, MIT (1977), Phil. Mag. 36/5, 1195-
1216 (1977).
167. A. S. Argon, M. M. Salama: Reports of Research in Mech. Process. of Polymers 22, MIT
(1977), Phil Mag. 36/5,1217-1234 (1977).
168. W. Retting: Die Angew. Makrom. Chemie 58/59, 133-174 (1977).
169. K. Smith, R. N. Haward: Polymer 18/7,745-746 (1977).
170. C. J. Singleton, E. Roche, P. H. Geil: J. Appl. Polymer Sci. 21,2319-2340 (1977).
171. R. P. Kambour: General Electric, Report No. 77CRD169, August 1977.
172. M. Ishikawa, I. Narisawa, H. Ogawa: J. Polymer Sci. (Polymer Phys. Ed.) 15, 1791-1804
(1977).
173. J. W. Sauer, E. Foden, D. R. Morrow: Polymer Eng. and Sci. 17/4, 246-250 (1977).
174. R. Bardenheier: DVM, Vortriige der 9. Sitzung des Arbeitskreises Bruchvorgiinge, 29-38,
11. 10. 1977.
175. J. Opfermann: DVM, Vortriige der 9. Sitzung des Arbeitskreises Bruchvorgiinge, 39-46,
11. 10. 1977.
176. K. P. Grosskurth: DVM, Vortriige der 9. Sitzung des Arbeitskreises Bruchvorgiinge, 47-60,
11. 10. 1977.
177. R. A. W. Fraser, I. M. Ward: Polymer 19/2,220 (1978).
318
178. J. A. Sauer: Polymer 19, NN (1978). S. Warty, D. R. Morrow, J. A. Sauer: Polymer, to be

published.
179. J.-C. Pollet, S. J. Burns: Int. J. of Fracture 13,667-679 (1977). J.-C. Pollet, S. J. Burns:
Int. J. Fracture 13/6, 775-786 (1977).
180. A. G. Atkins, C. S. Lee, R. M. Caddell: J. Mat. Sci. 10,1381 (1975).
181. K. B. Broberg: Arkiv for Fysik 18, 159-192 (1960).
182. G. T. Hahn, M. F. Kanninen: Fracture 1977, Vol. 1, ICF4, Waterloo, Canada, June 1977,
193.
183. K. Friedrich: Fracture 1977, Vol. 3, ICF4, Waterloo, Canada, June 1977, pg. 1119.
K. Friedrich: Dissertation, Ruhr-Universitat Bochum, 1978.
184. K. N. G. Fuller, P. G. Fox, J. E. Field: Proc. R. Soc. A(GB), VoL 341, No. 1627,537-557
(1974).
185. R. Weichert, K. Schonert: J. Phys. Mech. Sol. 22, 127-133 (1974).
186. R. Weichert: Dissertation, Universitat Karlsruhe, 1976.
187. P. C. Moon, R. E. Barker, Jr.: J. Polymer Sci (Polymer Phys. Ed.) II, 909-917 (1973).
188. E. Roeder, H.-A. Crostack: Kunststoffe 67/8, 454-456 (1977).
189. K. L. DeVries, D. K. Roylance, M. L. Williams: Report UTEC DO 68-056, Univ. of Utah,
Salt Lake City, 1968.
190. L. E. Nielsen, D. J. Dahm, P. A. Berger, V. S. Murty, J. L. Kardos: Polymer Sci. (Polymer
Phys. Ed.) 12, 1239 (1974).
191. M. D. Skibo, R. W. Hertzberg, J. A. Manson: J. Mat. Sci. 11, 479-490 (1976).
192. J. A. Manson, L. H. Sperling: Polymer Blends and Composites, New York/London 1976:
Plenum Press.
193. Mehrphasige Polymersysteme, Angew. Makromol. Chern. 58/59, 60/61 (1977).
194. J. Murray, D. Hull: J, Polymer Sci., Part A-2 8,583-594 (1970).
195. B. E. Nelson, D. T. Turner: Polymer Letters 9,677-680 (1970).
196. D. Hull, T. W. Owen: J. Polymer Sci. (Polymer Phys. Ed.) 11,2039-2055 (1973).
197. W. Doll: Colloid & Polymer Sci. 252, 880-885 (1974).
198. J. R. Martin, J. F. Johnson: J. Polymer Sci. (Polymer Phys. Ed.) 12,1081-1088 (1974).
199. J. Hoare, D. Hull: J. Materials Sci. 10, 1861-1870 (1975).
200. R. P. Kusy, D. T. Turner: Polymer 18, 391-399 (1977).
201. S. Gogolewski: Polymer 18,63-68 (1977).
202. S. Gogolewski, A. J. Pennings: Polymer 18,647-660 (1977).
203. J. E. Stamhuis, A. J. Pennings: Polymer 18,667-674 (1977).
204. S. Bandyopadhyay, H. R. Brown: Polymer 19.589-592 (1978).
205. E. S. Clark, C. A. Garber: Int. J. Polymeric Mater. 1, 31-46 (1971) ..
206. J. Patel, P. J. Philips: Polymer Letters Ed.11, 771-776 (1973).
207. E. Andrews: Pure and AppL Chern. 39/1-2,179-194 (1974).
208. G. Menges, N. Berndtsen: Kunststoffe 66/11, 735-740 (1976).
209. P. I. Vincent, S. Picknell, G. F. Harding: An Investigation of Fracture Criteria for Aniso-
tropic Polyethylene Terephthalate Film Using Mechanical and Optical Techniques, Div.
PoL Sci. Case Western Reserve Univ., Cleveland, Ohio 44106 (1967).
210. V. E. Korsukov, V. I. Vettegren', I. I. Novak, A. Chmel': Mekh. Polimerov 4,621-625
(1972). Polymer Mechanics 4, 536-539 (1972).
211. M. Takano, L. E. Nielsen: J. Appl. Polymer Sci. 20, 2193-2207 (1976).
212. H. Oberst: Kunststoffe 53, 4 (1963).
213. J. A. Sauer: J. Polymer Sci. C (Polymer Symposia) 32,69-122 (1971).
215. E. H. Andrews, B. J. Walker: Proc. R. Soc. Lond. A. 325,57-79 (1971).
216. J. P. Elinck, J. C. Bauwens, G. Homes: Int. J. Fracture Mech. 7, 277-287 (1971).
217. W. J. Plumbridge: J. Materials Sci. 7,939-962 (1972).
218. J. A. Manson, R. W. Hertzberg: CRC Crit. Reviews in MacromoL Sci. 1 (4), 433-499,
August 1973.
219. J. C. Radon, L. E. Culver: Polymer 16/7,539-544 (1975).
220. M. D. Skibo, R. W. Hertzberg, J. A. Manson: Fracture 1977, Vol. 4, ICF4, Waterloo,
Canada, June 1977,1127-1133.
319
221. N. K. Baramboirn: Mechanochemistry of Polymers (translated from the Russian), W. T.

Watson, ed., MacLaren 1964.
222. S. A. Komissarov, N. K. Baramboirn: Vysokomol. soyed. A11/5, 1050-1058 (1969).
Polymer Sci. USSR 11/5, 1189-1198 (1969).
223. N. K. Baramboirn, W. G. Protasow: die Technik 30/2, 73-81 (1975).
224. W. Lauer: Kautschuk + Gummi, Kunststoffe 28/9, 536-6l3 (1975).
225. A. Casale, R. S. Porter, J. F. Johnson: Rubber Chern. and Techn. 44/2, 534-577 (1971).
226. A. Casale: J. of Appl. Polymer Sci. 19, 1461-1473 (1975).
227. A. Casale, R. S. Porter: Polymer Stress Reactions, New York: Academic Press 1978.
228. T. Pazonyi, F. Tiidos, M. Dirnitrov: Plaste und Kautschuk 16/8, 577-581 (1969).
320
List of Tables
Table A. 1 List of Abbreviations of the most important Polymers. 321

Table A. 2 List of Abbreviations not Referring to Polymer Names 322
Table A. 3 List of Symbols . 323
Table A. 4 Conversion Factors . 325
Table A. 1. List of Abbreviations of the most important Polymers

(according to DIN 7723, DIN 7728, ASTM DI600-64T, ASTM 1418-67, ISO/R 1043-1969)
ABS Acrylonitrile-butadiene-styrene NBR Acrylonitrile-butadiene rubber

copolymer NCR Acrylonitrile-chloroprene
ABR Acrylonitrile-butadiene rubber
rubber NR Natural rubber
AMMA Acrylonitrile-methyl meth- PA Polyamide
acrylate copolymer PA6 Polycaprolactam
ASA Acrylester-styrene-acryloni- PA66 Po ly(hexamethy lene
trile copolymer adipamide)
BR Poly butadiene rubber PA 6.10 Po ly(hexamethy lene
CA Cellulose acetate sebacamide)
CAB Cellulose acetobutyrate PA 11 Polyamide of 11-amino-
CAP Cellulose acetopropionate undecanoic acid
CN Cellulose nitrate PA 12 Polylaurolactam
CP Cellulose propionate PAA Poly (acrylic acid)
CPE Chlorinated polyethylene PAN Polyacrylonitrile
CPVC Chlorinated poly(vinyl PB Polybutylene
chloride) (PBA) Poly butyl acrylate
EC Ethyl cellulose PBTP Poly(butylene glycol
EP Epoxide resin terephthalate)
EPDM Ethylene propylene terpolymer PC Polycarbonate
rubber PCTFE Po ly (chlorotrifluoroethy lene)
EPR Ethylene-propylene rubber PDAP Poly(diethyl phthalate)
EPTR Ethylene-propylene terpolymer PE Polyethylene
rubber PEC Chlorinated polyethylene
ETFE Ethy lene-tetrafluorethy lene PED Deuterated polyethylene
copolymer PEH PE (as opposed to PED)
EU Polyether-urethane PEO Polyethylene oxide
EVA Ethylene-vinyl-acetate PEOB (Poly[p-(2-hydroxyethoxy)
copolymer benzoic acid))
HDPE High-density polyethylene (PES) Polyethersuifone
HiPS High impact polystyrene PETP Poly( ethylene terephthalate)
IR Isoprene rubber PF Phenol-formaldehyde resin
IIR Butyl rubber, isoprene-iso- PI Poly-trans-isoprene
butylene copolymer PIB Polyiso buty lene
MF Melamin-formaldehyde resin PMMA Poly(methyl methacrylate)
321
List of Tables
Table A. 1. (continued)
PMP Poly( 4-methylpentene-l) PVDF Poly(vinylidene fluoride)

PaMS Poly-O!-methyl styrene PVF Poly(vinyl fluoride)
POM Polyoxymethylene =poly- PVFM Poly(vinyl formal)
acetal =polyformaldehyde py Unsaturated polyester resin
PP Polypropylene RF Resorcinol-formaldehyde resin
PPO Poly(2,6-dimethy1-1,4- RP Reinforced plastic
phenylene oxide) SAN Styrene-acrylonitrile
PS Polystyrene copolymer
PSU Polysuifone SB Styrene-butadiene copolymer
PTFE Poly(tetrafluoroethylene) SBR Styrene-butadiene rubber
PTMT Poly(tetramethylene SBS Styrene-butadiene block
terephthalate) copolymer
PUR Polyurethane SI Silicone rubber
PVA(L) Poly(vinyl alcohol) SIR Styrene-isoprene rubber
PVAC Poly(vinyl acetate) SMS Styrene-<l'-methy lstyrene
PVB Poly(vinyl butyral) copolymer
PVC Poly(vinyl chloride) UE Polyurethane rubber
PVCA Vinyl chloride-vinyl acetate UF Urea-formaldehyde resin
copolymer UP Unsaturated polyester resin
PVDC Poly(vinylidene chloride) UR Polyurethane rubber
Table A. 2. List of Abbreviations not Referring to Polymer Names
CN Center notched specimen (fracture mechanics)

COD Crack opening displacement (fracture mechanics)
CT Compact tensile specimen (fracture mechanics)
DCB Double cantilever beam (fracture mechanics)
DT Double torsion (fracture mechanics)
ESR Electron spin resonance
IR Infra red
NMR Nuclear magnetic resonance
SANS Small angle neutron scattering
SAXS Small angle X-ray scattering
SEM Scanning electron microscopy
SEN Single edge notch specimen (fracture mechanics)
UV Ultra violet
WAXS Wide angle X-ray scattering
WOL Wedge open loaded specimen (fracture mechanics)
322
List of Tables
Table A. 3. List of Symbols

Symbol General Meaning (if not indicated otherwise Unit Used
in the section where the symbol appears)
C Concentration of end groups molm-3
C Bending compliance mN-I
Cp Specific heat J kg-I K- I
D Bond dissociation energy kJ mol- I
D Average spherulite diameter J,lm
E Young's modulus GNm- 2
E' Storage modulus GNm-2
E" Loss modulus GNm- 2
E* Complex modulus GNm- 2
EA Activation energy kJ mol-lor
kcal mol- I
Ec Elastic modulus of crystalline regions GNm- 2
Ek Young's modulus of chains GNm- 2
Et Creep modulus GNm- 2
.rlF Activation energy kJ mol- I
G Shear modulus GNm-2
G' Storage modulus Nm- 2
G" Loss modulus Nm-2
G* Free enthalpy of activation kcal mol- I
GA Energy release rate at crack arrest Jm-2
Gc Energy release rate corresponding to Kc Jm-2
Gel Plain-strain energy release rate Jm-2
Gc2 Plain-stress energy release rate Jm-2
(plastic component of Gc)
Gd Dynamic energy release rate Jm-2
Glc Critical energy release rate Jm-2
K Compressive modulus GNm-2
Ko Initial stress intensity factor MN m- 3/ 2
Ke Fracture toughness - the largest value of MN m-3/ 2
stress-intensity factor that exists prior
to onset of rapid fracture
Kel Plane strain fracture toughness MN m-3/ 2
Ke2 Plane stress fracture toughness
KJ Stress intensity factor MNm- 3/ 2
Km Stress intenstiy factor below which no MN m-3/ 2
crazing occurs
L Contour length of a segment nmor A
Lo End-to-end distance of a segment nmorA
M Molecular weight (g mol-I)
Mc Chain molecular weight between (g mol-I)
crosslinks
Mn Number average molecular weight (g mol-I)
Mw Weight average molecular weight (g mol-I)
N Number of chains between cross links
NF Number of cycles to failure
NL Avogadro number 6.022 . 10 23 mol- I
Qb Amount of heat dissipated kJ
R Molar gas constant 8.314 J mol- I K- I
S Internal entropy kJ K-I
Sy Volume of liquid absorbed/unit voL
of polymer
323
List of Tables
Table A. 3. (continued)
Symbol General Meaning (if not indicated otherwise Unit Used

in the section where the symbol appears)
T Absolute temperature K
Tg Glass transition temperature °c
U Internal energy kJ mol- 1
Uo Activation energy for bond scission kJ mol-lor
kcal mol- 1
a Crack length m
an Impact strength kJm-2
aT Shift factors of relaxation times
f Fractional unoccupied volume cm 3
g Spectroscopic splitting factor
h Planck's constant 6.626 . 10-34 J s
k Boltzmann's constant 1.380· 10-23 JK- 1
kc Rate of thermo mechanical chain scission s-1
kS Notch sensitivity factor of strength
kT Notch sensitivity factor of energy to break
mI Spin quantum number
n Number of chain atoms
~ Birefringence
Pc Craze yield stress MNm-2
q Molecular cross-section m2
r Chain end-to-end distance nmorA
rp Plastic zone size Ilm
tb Time to break s
0< Activation volume m 3 mol-lor nm 3
(3 Activation volume for chain scission m3 mol-lor nm 3
'Y Activation volume of fracture under m 3 mol-lor nm 3
tensile stress
/j Solubility parameter cal 1/ 2 cm- 3/ 2
tan Ii Loss tangent
€ Strain
€c Critical strain
€y Yield strain
1) Viscosity (dynamic) Nsm- 2
A Uniaxial extension ratio
A Heat conductivity WK- 1 m- 1
As Linear swelling ratio
Il Coefficient of friction
v Poisson's ratio
P Polymer density g. cm- 3
Pa Radius of crack tip Ilm
U Macroscopic stress GNm- 2
ub Breaking strength GNm-2
Uc Critical stress GNm-2
r Relaxation time s
r* Critical shear stress GNm-2
Axial chain stress GNm- 2
'"'"c Chain strength GNm-2
w Loading frequency Hz
w Angular frequency Hz
324
List of Tables
Table A. 4. Conversion Factors
Frequently found SI-Units Conversion

non-SI Units
atmosphere, at Nm-2 or Pascal, Pa 1 at = 98.07 . 10 3 N m-2

bar 1 at =0.9807 bar
calorie, cal Joule, J 1 cal =4.187 J
centipoise, cP Pascal second, Pa· s 1 cP = 1.000 . 10-3 Pa . s
dyn Newton, N 1 dyn =1 . 10-5 N
erg Joule, J 1 erg = 10-7 J
foot, ft meter, m 1 ft =3.048.10-1 m
foot pound-force, lbft Joule, J 1 Ib.ft = 1.356 J
inch, in meter, m 1 in = 2.540 . 10-2 m
kilopond,kp Newton,N 1 kp= 9.807 N
kilogram-force per square NM-2 or Pascal, Pa 1 kgf cm-2 = 9.807 . 104 N m- 2
centimeter
mil (= 0.001 in) micrometer, /.Lm 1 mil = 25.4 /.Lm
pound per square inch, PSI Nm-2 or Pascal, Pa 1 PSI = 6.897 . 10 3 N m- 2
325
Subject Index
Key words and page numbers set in italics refer to the headings.
Accumulation of defects 64 Breaking kinetics of fibers 44

Acoustic emission 288 Brinell hardness 57
Activation energy 55 Brittle fracture, PVC 2
for main chain scission of radicalized Brittle point temperature 58
chains 85 Buckling of microfibrils 202
(see also bond energy and dissociation Bundle of chains 36
energy)
for chain translation 105 Carbanions 170
Activation volume 55 Carbon black 237
- for bond scission 112 Catalyst 20
- of creep failure 217 Cavity formation 10, 236, 291
Adhesive joint 51 Chain conformation 89
Adjacent reentry 19 Chain end separation 88
Affine deformation 33, 39 Chain entanglement 218,283,287,311
Aggregate model 33 Chain folding 18,160
Aggregates Chain mobility 291
of chains 32 Chain molecule 87
cylindrical 21 characteristic properties 5
globular 20 conformation 16
spherical 21 mechanical properties 6
Amorphous state 15, 27 orientation 21,59, 217
- model representations 16 (see also draw ratio)
Andrade equation 214 in solution 109
Anelastic deformation 25, 27 Chain scission 10,113, 141, 173, 287
Anharmonic oscillator 86 (see also radical formation)
Anisotropy of the spectrum 142 Chain segments
Anomalous decay 170 ~- basic transformations 10
Apparent energy of bond scission 152 Chain slip 108
Arrhenius equation 216 Chain strength 109, 141
Atomic orbitals 69 - mechanical properties 11
Axial compression of an extended chain 97 Chain stress 114
Chain tension
Bead spring model of Rouse 109 - in PA-6 crystals 104
Biaxial yielding 49 - in PE crystal 102
Binding potential form 86 Change of conformation 10, 111
Biological degradation 237 Characteristic ratio 88,91
Blends 21, 30, 2~0 Characterization 23
Block copolymers 21, 290 Charpy test 204
Boltzmann factor 113 Classical failure criteria 47
Boltzmann relation 89 Clo se range order 15
Boltzmann superposition principle 26,229 Coating of the surface 293
Bond Cohesive energy 101, 103
density 57 Color centers 233
energies 73,152 Comminution 313
orbitals 72 Complex modulus 26
rotation 88 Complex shear modulus of a composite 30
strength 80 Composites
Breakage of primary bonds 56 - complex modulus 30
326
Subject Index
polymer-polymer see Blends, Defects 219, 229, 282

see also Polymer blends, Block accumulation 61
copolymers, - development 9
Graft copolymers - migration 160
Compression 97 Deformation 24, 174, 204
Conformation 5,16,89 - molecular description 27
Connectivity 57 Degradation 128,237
Continuous viscoelastic models 52 Dichroism 176
Continuum mechanics 47 Difference IR spectrum 185
Cooperative motion 114 Disclination 36
Copolymer Disentanglement of molecular coils 218,
- butadien-styrene diblock 22 287,312
Coulomb yield criterion 48 Dislocations 27,36,105
Covalent bonds 69 Dissipated heat 288
Crack 251 Dissociation energy 79
arrestor 166 - stress effect 114
initiation 207 Distribution of tie-segment lengths 147,
opening displacement 254, 270 189
propagation Double beam technique 183
molecular and morphological Drawing 230
aspects 293 Draw ratio 33, 211, 217
transpherulitic 312 Ductile-brittle transition 59
- speed 267 Ductile deformation 27
Crazes 164, 272 Ductile failure
initiation 276 - HDPE 4
propagation and breakdown 286 - PVC 2
response to environment 290 Dugdale model 257
shape 286 Dynamic loading of a chain 108
suppression mechanism 293 Dynamic strain energy release 271
Creep crack 3 Dynamic tensile modulus
- in LDPE tubes 215 - polymer blend 30
Creep craze
- HDPE 4
Creep failure 212 Elastic modulus 6,93
- kinetic parameters 216 - PE single crystal 18
Creep strain Elastomers 235
- PVC 213 Electron affinity 84
Criegee mechanism 238 Electron density 69
Critical energy release rates 271 Electronic excitation 81
Critical tensile strength 6, 58 Electronic energy 73
Cross-link 235 Electron microscopy 23
Cross-link density 165, 211 Electron paramagnetic resonance (EPR) 120
Cross-linking agent 165 Electron-scattering 23
Crystal distortion 183 Electron spin resonance (ESR) 120
Crystal lattices Elliptical hole 50
- elastic interaction with displaced End group formation
chains 99 - kinetic parameters 187
Crystallites 1 7 Energy elastic chain deformation 93, 236
- extended chain 21 Energy of dissociation 73
- plastic deformation 27 Energy of distortion 48
Crystallization Energy release in chain scission 196
solidification-model 19, 294 Energy release rate 254, 260
- potential 99 Entanglement molecular weight 283
Enthalpy
Damage accumulation 199 atomization of hydrogen 78
Debonding 51 - sublimation of graphite 78
327
Subject Index
Enthalpy relaxation 111 patterns 2

Entropy 197 surfaces 66,293
Entropy-elastic deformation 88, 236 HDPE under impact 207
Entropy relaxation 111 polyamide 149
Entropy of a single chain 89 polymethylmethacrylate 301
Environmental degradation 237 polypropylene 298
EPR 119 theories 40
Equivalent random link 89 - rate process 53
ESR investigations 141 under hydrostatic pressure 50
neck formation in polycarbonate 232 Fracture Mechanics 47,50,252
spectra 119 - of viscoelastic media 51
assignment 128 Fracture toughness 256, 260
- hyperfine structure 121 Free energy 87
- of 6 polyamide fibers 144 Freely jointed chain 88
Excited states 80 Free radicals 119, 229
Extended-chain 160 amorphous regions 142
- crystals 22, 295 concentration 135
- stretching 97 formation 141
Extended conformation 147 identification of spectra 123
Extended length 88 in ground high polymers 125
Eyring theory of flow 54 in stressed rubbers 162
in tensile specimens 126
in unsaturated rubbers 238
Failure envelope 52, 236 number of spins 122
Failure surface 47 standard sample 123
Failure under constant load 211 Frequency sensitivity
Falling-weight test 204 - fatigue crack growth 311
Fatigue 171, 220 Friction 108
- cracks 310 Fringe micelle 17
- fracture of fibers 198
Fiber strength 6
Gaussian distribution 46
- effect of molecular weight 12
Gaussian theory 89
- effect of specimen length 45
Glass beads 166
Fibrillar structure 28
Globular structure 193
Fibrillation 201
Graft copolymers 21, 30
Filler Material 237
Griffith fracture criterion 50, 238, 252, 259
Filler particle 235
Griffith theory 50
Fillers 165
Fire and heat stability 237
Flaw 44,47 Hamiltonian operator 70
- rotation 51 Heat of formation 73
Forces of inertia 111 Heat treatment
Formation of microcracks 193 - effect on structure 159
Forming of polymeric materials 235 Hesitation lines 302
Fractography 200, 293 Heterogeneous fracture 251
Fracture criterion 252, 258 Histograms of free radical concentration
classical 47 148, 159
concentration of end groups 187 Hole formation 235
critical energy release rate 260, 271 Homogeneous fatigue 220
fracture toughness 256, 260 Homogeneity of segment lengths
Fracture - uniaxial strength 191
definition 9 Homolytic bond scission 84,112
initiation 64 Hooke's law 24
kinetic theory 27 Humidity 237
multiaxial stress 49 Hybridization 71
of elastomers 65 Hybrid orbitals 71
328
Subject Index
Hydrocarbon radicals 121 - PE 4,215

Hydrogen bonding parameter 292 - PVC 2,213
Hydrogen bonds 241 Low temperature relaxation mechanisms
Hydrogen-like atom 70 27,283
Hysteresis heating 222, 312
Mackerel patterns 300

Impact fracture 306 Magnetic moment 120
- unoriented polymers 204 Many-electron atom 70
Impact resistance rating 205 Mastication 313
Impact strength 205 Material inhomogeneities 44
- degree of cross-linking 211 Matrix orientation 51
Impurities 127, 165 (seel also orientation)
Inclusions 49 Maximum elastic strain 48
Incompatibility 21 Maximum principle stress 47
Infra-red absorption 112 Maxwell element 26
- end group concentration 183,305 Meander model 34
- stress effect of IR bands 114, 177 Mechano-chemistry 312
Infra-red technique 224 Mechano-radicals 141
Inhomogeneous blending 219 Mechanical excitation 99
Initiation of failure 49, 173 Mechanical degradation 157
Interfibrillar cohesion 202 Mechanical properties
Intermittent load application 240 - effect of orientation 60
Intermolecular attraction 57 Mechano-chemical reaction chain 194
Internal energy 197 Melt 36
Internal entropy production 87 Microcracks 242
Internal rotation 88 - formation 193,250
Intramolecular binding potential 74,81, 115 Microfibers 2, 164
Intrinsic binding energy 73 Micromorphology of the craze 274
Intrinsic flaw sizes 268 Microstructure 15
Intrinsic viscosity 109 - (see also crystallite, spherulite, super-
Ionic dissociation 84, 112, 133 structure)
Ionization 82 Microvoids 229
- potential 84 Microwave energy 120
IR see infra-red Miner's rule 199
Izod test 204 Miscibility 21
Model of craze initiation 284
Model representation of deformation 29
Kinetic theory of fracture 56, 217, 270
Models of the structure
Kinkenmodell 34
- PE, PA 6 28
Kink-isomers 92
Modes of controlled crack propagation 252
Molecular anisotropy 31
Langevin function 90 Molecular friction coefficient 109
Laplacian distribution 46 Molecular orbitals 72
Laundering detergent 241 Molecular orientation, effect on impact
Life expectancy 43 strength 211
Light-scattering 16, 23 (see also chain orientation and orienta-
Limiting molecular weight 110 tion effect)
Line profile measurements 34 Molecular packing 224
Liquid gases 292 Molecular weight 250
Logarithmic normal distribution 41 effect on fatigue 230
Long-time failure 211 effect on strength 11, 59
eN 57 effect on tensile strength of polyamide
model network 65 6, 12
PA 57 - effect on viscosity in solution 109
PAN 63 Mooney-Rivlin plot 239
329
Subject Index
Morse potential 86, 114 Polymers

Mosaic-block structure 171, 292 hard elastic 23
Mullins effect 235 - list of standardized abbreviations 321
Multi-phase polymers 21 - ultra oriented 23
Multiple bonds 73 Polymer-solvent interaction 291
Preorientation 162
Nascent polymer structure 20 Primary bonds 69
Necking 230 . Primary radicals 129
Neutron scattering 16,19,23 Probability of fracture 46
Nitric acid etching 171 Propagating stress waves 111
Non-equilibrium response 98
Non-linear viscoelasticity 213 Quantum chemistry 69
Nonrandom scission 250
Notch sensitivity 306 Radical
Nuclear magnetic resonance (NMR) 23 decay 168
ion 84
concentration
Octahedral shearing stress 48
- fracture surfaces 157
One-electron wave function 69
production
Optical microscopy 23
concentration at break 156
Orientation behavior of amorphous thermo-
effect of strain rate 150
plastics (PS, PMMA, PVC) 33
effect of temperature 152
Orientation effect
rate of bond scission 154
on fracture strength 51,57,63,217
reactions 141
on free radical formation 162
- mechanical relevance 167
on hyperfine structure 122
recombination 168
on impact strength 211
scavengers 220
on Young's modulus 33
transfer 168
Oscillator 112
trapping sites 127, 170
Oxidative degradation 239
Ozone 240
Radicalization 84
- degradation 238
Radicalized chains 80
Radius of gyration 16, 19
Rain erosion 237
Paired electron spins 70 Random coil structure 16,89
Parabolic markings 302 Random copolymers 21
Paracrystal 34 Rankine criterion 220
Partition function 35,92, 113
Rate of dissociation 113
Pauli principle 70, 120
Reaction enthalpy 114
Phase separation 109
Reaction rate theory 53, 212
Phenomenological theory of viscoelasticity
Relaxation maxima 269
26
Relaxation time spectra 26, 110
Photo-chemical degradation 241
Repeated straining 190
Photon absorption 82
Reuss model 33
pH-values
Rigidity of the crystal blocks 145
- effect on the fatigue life 241
Rod model 31, 60
Plastic deformation 25,256,293, 306
Root-mean-square length 88
Plasticization 240,293
Rotational isomers 92
Poisson distribution 66
Row structures 22
Poly-
Rubber-toughened plastics 290
In view of the very large number of
citations polymer names are not included
in this subject index; please refer to the Schrodinger equation 69, 86
property or phenomenon of interest. Scission 111
Poly-blends 21 (see also radical formation)
Polymer fatigue 220,310 Secondary radicals 129
- molecular interpretations 224 Segment homogeneity 162
330
Subject Index
Semicrystalline polymers 28 Takayanagi diagram 29

Shear degradation 109 Tensile impact 204
Shear yielding Tensile mechanical loss tangent
- molecular model 231 - PC 228
Shearing forces 109 - PS 226
Shift factor Tensile strength 173,230
- related to cross-link density 240 - morphology 13
Shish-kebabs 22 Thermal crack growth 64,207
short-range order Thermal degradation 240
(see close range order) Thermal fatigue failure 222
Silica 166, 237 Thermal measurements 288
Single crystals 27 Tie molecule 100, 145
- PE 17 Time-dependent strength of a chain segment
Skeleton vibrations 114 112
Slip 10 Time-temperature diagram
Slipping of secondary bonds 55 - PVC 8
Small angle X-ray scattering 16, 34, 229 Time-temperature superposition principle 53
Small-strain deformation 174 Time to fracture 57,62
Solid state extrusion 22 effect of orientation 63
Solubility parameters 291 - of PE tubes 4,42
Solvent crazing 291 - of PVC tubes 3
Spectroscopic splitting factor 120 Traction vector 32
Spherical inclusion 30 Triblock copolymers
Spherulite 18,292 - styrene-butadiene 166
Splitting of fibers 201 Triboelectric charges 170
Statistical aspects of fracture 41 Twisted translation 100
Statistically kinked chain 88 Twisting lamellae 18
Statistical theory of extreme values 45
Step-temperature test
- 6 polyamide 153 Ultra high modulus fibers 234
Stiffness 6,95 Ultra high molecular weight 295
Stochastic variable 41 Ultraso nic d egradatio n 110
Strength 6 Underlying distribution 46
- molecular theory 47 Uniaxial creep 214
Stress concentration 51, 252, 307 Uniaxial deformation 173
Stress cracking of polyamides - at high rates 111
- metal salts 292 - elastomers 25
(see also crazes) - thermoplastic polymers 24
Stress intensity factor 253 Unoccupied volume 210
Stress relaxation 111, 146, 174, 188 Unpaired free electrons 119
Stress space 49 UV irradiation 241
Stress-strain curve 87
of craze material 276 Variability 42, 43
- of polyamide 6 chain segments 115 Viscoelastic behavior 23, 256
relation to radical Viscoelastic extensibility 65
concentration 188 Voigt-Kelvin element 26
Stress tensor 32 Voigt model 33
Stress whitening 213 Von Mises yield criterion 220, 257
String-model 161
Striations 164
Structure sensitive kinetic parameter 63, 216 Wallner lines 302
Su bmicrocracks 193 Wavefronts 294
Subcritical crack growth 259 Weathering 241
Superstructure 15 Weibull distribution 46
Surface coating 312 Whitening 234
Swelling 240, 290 Wohler curves 223
331
Subject Index
X-ray diffraction 23 Yield stress 231

X-ray scattering 224, 229, 242 Young's modulus 58
Yielding 230
Yield point 24 Zero point energy 73, 78
332

Polymers

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Polymers

Uploaded by

Copyright:

Available Formats

Polymers

Properties and Applications 2

Laboratoire des Polymeres, Departement des Materiaux,

ISBN-13: 978-3-642-96462-6 e-ISBN-13: 978-3-642-96460-2

Library of Congress Cataloging in Publication Data:

© by Springer-Verlag Berlin Heidelberg 1978

Lausanne, September 1978 H. H. Kausch

Chapter 1 Deformation and Fracture of High Polymers,

Chapter 2 Structnre and Deformation 15

Chapter 3 Statistical, Continuum Mechanical, and Rate Process

Chapter 4 Strength of Primary Bonds 69

II. Bond Energies . . . . . . . . . . . . . . . 73

Chapter 5 Mechanical Excitation and Scission of a Chain 87

Chapter 6 Identification of ESR Spectra of Mechanically Formed Free

Chapter 7 Phenomenology of Free Radical Formation and of Relevant

3. Concentration at Break 156

Chapter 8 The Role of Chain Scission in Homogenous

Chapter 9 Molecular Chains in Heterogeneous Fracture 251

III. Molecular and Morphological Aspects in Crack Propagation 293

Subject Index. . . . . . . . . . . 326

Deformation and Fracture of High Polymers,

The importance of a thorough understanding of the deformation behavior and the

Fig. 1. 3. Long-time failure of a PVC pipe through

Fig. 1.6. Ductile failure of a HOPE water pipe under

bond +---- segment

Fig. 1.9. Representation of

tation of a (polyamide 6) chain showing the nonextended collformation, an

1. Polyethylene PE 28 235,000 "'0 "

4. Poly(oxymethylene) POM 1.92

9. Polystyrene PS 104 2.21 9000-11 ,800 3250 50

Chain molecule Oriented Fiber [17] Isotropic Solid

1043- weight L(lO-lOm) Ec(MN/m2) E ub E ub Uc 8'

Natural rubber NR 136 8.1 1-2 17-25

Fig. 1.10. Arbitrary segment of polyamide 6

namely the anisotropy of macroscopic properties, the microscopic heterogeneity

200 100 50 o -50 e

transformation of pairs excess volume molecular process

the fracture process is understood as stress-biased material disintegration. It carries

0.8 // x/, m 140 •C

Fig. 1.14. Tensile strengths of polyamide 6 at different levels of morphologic organization.

continuum viewpoint. The same approach is applied to the phenomenon of craz-

References for Chapter 1

1. H. A. Stuart (ed.): Physik der Hochpolymeren. Berlin - Heidelberg: Springer-Verlag 1952.

5. I. M. Ward: Mechanical Properties of Solid Polymers. London: Wiley-Inter science 1971.

Structure and Deformation

I. Elements of the Superstructure of Solid Polymers . 15

I. Elements of the Superstructure of Solid Polymers

Fig. 2.3. Chain folding in polyethylene single crystals.

At this point the superstructure or morphology of polymers is reviewed with a desire

Presently it is discussed [91,92,100] whether and to what extent the lamellar

Fig. 2.8. Polypropylene prepared with

A large group of multi-phase polymeric systems is formed by the block co-

The experimental methods available to determine or characterize the structure of

Even at these small deformations the apparent Young's modulus E is a function

b) Hard elastic PP/

The shear modulus G is proportional to the absolute temperature T, Boltzmann

G = NkT = pRT (1-2 Me/M)/Me (2.3)

where M and Me are the number-average molecular weights of the uncrosslinked

u~ U crystall ine region

III. Model Representation of Deformation

For the interpretation of the complex mechanical response of a highly anisotropic

Fig. 2.12. Takayanagi diagram of semicrystalline polymer; con-

meaningful interpretation of the relative contributions of crystalline and amorphous

The quantity a is derived from the Poisson's ratio vm as: